A Network-Based Method to Analyze EMI Events of On-Board Signaling System in Railway

The On-board Train Control System (OTCS) plays an important role in the real-time operation of the electric multiple units (EMU) in high-speed railway. The EMU is a complex system made up of various electrical and electronic equipment, so the interactions of the electromagnetic (EM) environment and OTCS are difficult to study, which leads to more challenges to analyze EM interference (EMI) events at the system level. To overcome this difficulty, this paper proposes the thought of a graph model to solve the problem. First, a framework is proposed to more clearly reflect the relationship between the EMC (Electromagnetic Compatibility) problem and network through a comparison with them. Second, a network theory-based model is presented to express the EMC elements for the OTCS in EMU. It decomposes the OTCS and EMU with EMC elements into edges and nodes of the network, which parameters are defined corresponding to EM sources, sensitive equipment, and coupling paths. Thus, each part could be modeled separately or together by calculation, simulation, or measurement, respectively, and the EMC problem could be represented by the paths from origin to destination in the network. Moreover, the modeling process was elucidated by the specific cases in OTCS and the validity of the proposed approach was verified by calculation and measurement results in the case study.


Introduction
The electric multiple units (EMU) are becoming increasingly advanced and intelligent with the new technology applied in the world railway system, such as the new types of CR (Fuxing in China Railway) series in China in the last years. Risks from electromagnetic (EM) disturbance to the On-board Train Control System (OTCS) has been spotlighted and investigated in the railway industry [1][2][3][4][5], because it may cause serious fault or failure when the electric and electronic parts are subjected to them. To ensure the safe operation of the EMU, system-level EMC analysis of OTCS is essential and it is regarded as one tough work with the following features. The OTCS is a complex system that the subsystems and components are distributed throughout the EMU. During the continuous movement of EMU, interactions between high power sources (mainly from the traction power supply and drive system) and safety-critical systems (OTCS) will occur, and it is a different EM environment from other industries. Thus, EM modeling and test for every EMC problem of subsystem require large-scale computation and massive measurement. However, the traditional solutions facing such a system-level assessment are limited.

EM Interactions between EMU and OTCS
For the EM environment in high-speed railway, different coupling manners are connecting the EM disturbance sources of the EMU and various parts of the OTCS equipment. The relations between the source in EMU and the equipment at the dispersed position are shown in Figure 3 and various EM couplings make the interactions extremely complicated. In general, there are two types of EM coupling, namely, conductive coupling and radiation coupling. The former includes conductive direct coupling and capacitive and inductive coupling. The latter contains antenna coupling, field line coupling, and cable coupling. Different sources can affect the target devices through the aforementioned coupling methods. As shown in Figure 3, the EMU can be divided into the roof, interior, and bottom in space. Correspondingly, OTCS equipment is also distributed in the above three spaces, such as cables, antennas, cabinets, grounding, and sensors. Therefore, each EMI event corresponds to an interactive relationship in the figure. Thus, all events will become a complex network of connections. It will be helpful to analyze the OTCS EMC problems if all the interactions are transformed into the form of the network.

EM Interactions between EMU and OTCS
For the EM environment in high-speed railway, different coupling manners are connecting the EM disturbance sources of the EMU and various parts of the OTCS equipment. The relations between the source in EMU and the equipment at the dispersed position are shown in Figure 3 and various EM couplings make the interactions extremely complicated. In general, there are two types of EM coupling, namely, conductive coupling and radiation coupling. The former includes conductive direct coupling and capacitive and inductive coupling. The latter contains antenna coupling, field line coupling, and cable coupling. Different sources can affect the target devices through the aforementioned coupling methods.

EM Interactions between EMU and OTCS
For the EM environment in high-speed railway, different coupling manners are connecting the EM disturbance sources of the EMU and various parts of the OTCS equipment. The relations between the source in EMU and the equipment at the dispersed position are shown in Figure 3 and various EM couplings make the interactions extremely complicated. In general, there are two types of EM coupling, namely, conductive coupling and radiation coupling. The former includes conductive direct coupling and capacitive and inductive coupling. The latter contains antenna coupling, field line coupling, and cable coupling. Different sources can affect the target devices through the aforementioned coupling methods. As shown in Figure 3, the EMU can be divided into the roof, interior, and bottom in space. Correspondingly, OTCS equipment is also distributed in the above three spaces, such as cables, antennas, cabinets, grounding, and sensors. Therefore, each EMI event corresponds to an interactive relationship in the figure. Thus, all events will become a complex network of connections. It will be helpful to analyze the OTCS EMC problems if all the interactions are transformed into the form of the network. As shown in Figure 3, the EMU can be divided into the roof, interior, and bottom in space. Correspondingly, OTCS equipment is also distributed in the above three spaces, such as cables, antennas, cabinets, grounding, and sensors. Therefore, each EMI event corresponds to an interactive relationship in the figure. Thus, all events will become a complex network of connections. It will be helpful to analyze the OTCS EMC problems if all the interactions are transformed into the form of the network.

Framework of the EMC Network Model
The framework of the EMC network model is composed of two parts with a pink and blue dashed box, as shown in Figure 4. In the first part, the EMC network is proposed based on engineering and historical cases. Due to the complexity of the disturbance source and coupling mechanism, we have to investigate all possible situations under consideration. Thus, it is necessary to have comprehensive research on the characteristics of the disturbance such as location, frequency, types of coupling, and so on. In the second part, to analyze the EMI events, the structure and function of the OTCS should be known especially the connection and spatial relationship between the OTCS and EMU. Moreover, the modeling process is consistent with the interaction between the EM environment and system structure. The final EMC network model is based on the modeling process of this framework and the model consists well with the principle of the EMI event, i.e., the disturbance generated in the EM environment propagates in EMU and ultimately interferes with the OTCS in high-speed railway. However, the network model is only an auxiliary methodology that can help us to make the EMC analysis of the system more comprehensive. The study of each EMI event still relies on traditional EMC analysis methods, and it will be explained in the case study. In the following parts, the model is named the EN model for simplicity.

Framework of the EMC Network Model
The framework of the EMC network model is composed of two parts with a pink and blue dashed box, as shown in Figure 4. In the first part, the EMC network is proposed based on engineering and historical cases. Due to the complexity of the disturbance source and coupling mechanism, we have to investigate all possible situations under consideration. Thus, it is necessary to have comprehensive research on the characteristics of the disturbance such as location, frequency, types of coupling, and so on. In the second part, to analyze the EMI events, the structure and function of the OTCS should be known especially the connection and spatial relationship between the OTCS and EMU. Moreover, the modeling process is consistent with the interaction between the EM environment and system structure. The final EMC network model is based on the modeling process of this framework and the model consists well with the principle of the EMI event, i.e., the disturbance generated in the EM environment propagates in EMU and ultimately interferes with the OTCS in high-speed railway. However, the network model is only an auxiliary methodology that can help us to make the EMC analysis of the system more comprehensive. The study of each EMI event still relies on traditional EMC analysis methods, and it will be explained in the case study. In the following parts, the model is named the EN model for simplicity.

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1.

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows: 1.
The definition and representation of the node The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1.

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1.

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1.

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2.

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the results of the network achieve an acceptable level, the process is stopped and the model is analyzed in the next step. Otherwise, the model should be simplified by several times until there are no repetition and conflict between the nodes and edges. After the work above, the expression can be given with the format as: where EN denotes the EN model; Av and Ae are the attribute function for the node and edge separately, thus Av and Ae can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2.

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the results of the network achieve an acceptable level, the process is stopped and the model is analyzed in the next step. Otherwise, the model should be simplified by several times until there are no repetition and conflict between the nodes and edges. After the work above, the expression can be given with the format as: where EN denotes the EN model; Av and Ae are the attribute function for the node and edge separately, thus Av and Ae can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured on-site. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

2.
The definition and form of the edge Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2. There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2.
where EN denotes the EN model; Av and Ae are the attribute function for the node and edge separately, thus Av and Ae can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2.

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the results of the network achieve an acceptable level, the process is stopped and the model is analyzed in the next step. Otherwise, the model should be simplified by several times until there are no repetition and conflict between the nodes and edges. After the work above, the expression can be given with the format as: where EN denotes the EN model; Av and Ae are the attribute function for the node and edge separately, thus Av and Ae can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2.

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the results of the network achieve an acceptable level, the process is stopped and the model is analyzed in the next step. Otherwise, the model should be simplified by several times until there are no repetition and conflict between the nodes and edges. After the work above, the expression can be given with the format as: where EN denotes the EN model; Av and Ae are the attribute function for the node and edge separately, thus Av and Ae can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2.

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the results of the network achieve an acceptable level, the process is stopped and the model is analyzed in the next step. Otherwise, the model should be simplified by several times until there are no repetition and conflict between the nodes and edges. After the work above, the expression can be given with the format as: where EN denotes the EN model; Av and Ae are the attribute function for the node and edge separately, thus Av and Ae can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2.
where EN denotes the EN model; Av and Ae are the attribute function for the node and edge separately, thus Av and Ae can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: Wireless transmission 1 The edge between the source and coupling nodes are the same with coupling connection. 2 Field-line coupling and inductive or capacitive coupling (crosstalk) are classified as this form for simplicity.

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the results of the network achieve an acceptable level, the process is stopped and the model is analyzed in the next step. Otherwise, the model should be simplified by several times until there are no repetition and conflict between the nodes and edges. After the work above, the expression can be given with the format as: Appl. Sci. 2020, 10, 9059 where EN denotes the EN model; A v and A e are the attribute function for the node and edge separately, thus A v and A e can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: (1) Basic information and materials to build the EN model are obtained from the system structure, documents, historical cases, expert experience, on-site measurement, and so on. (2) The nodes in the EMC domain and system structure domain are achieved and the properties are waiting for the decision by proper method. (3) According to the actual spatial location and connection relationship, the edges between the nodes are determined by the definition in Table 2. (4) The preliminary EN model will be adjusting, merging, and deleting nodes and edges for simplification. Then, the attribute function of the nodes and edges could be got by calculations, simulations, and measurements on the basis of their properties. (5) Repeat the steps above, and the final EN model is ready for application after optimization.
Each path in the model can present a specific EMI event.
After that, the EMC problems of the target system are presented via a network based on the EN model.

Traction System and the OTCS in EMU
The OTCS equipment is one of the important technical equipment to ensure the safe operation of the EMU in high-speed railway. The EMC problem in OTCS is mainly caused by the traction system in EMU, so the traction system and OTCS are first analyzed to obtain the information and materials and how to model.

Traction System
The traction system of the EMU consisted of the pantograph, high-voltage cable, traction transformer, traction converter, auxiliary power unit, traction motors, and ground device as shown in Figure 5. From the physical and electrical structure, it can be known that most of the devices and components of the traction system are concentrated on the top and bottom space of the EMU, except the pantograph on the roof. The electric energy is obtained through sliding on the contact wire of the catenary by the pantograph, and the motors propel the EMU in the traction drive system. As the conclusions in References [4,8], pantograph arcing is one of the sources in EMU.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 18 (1) Basic information and materials to build the EN model are obtained from the system structure, documents, historical cases, expert experience, on-site measurement, and so on. (2) The nodes in the EMC domain and system structure domain are achieved and the properties are waiting for the decision by proper method. (3) According to the actual spatial location and connection relationship, the edges between the nodes are determined by the definition in Table 2. (4) The preliminary EN model will be adjusting, merging, and deleting nodes and edges for simplification. Then, the attribute function of the nodes and edges could be got by calculations, simulations, and measurements on the basis of their properties. (5) Repeat the steps above, and the final EN model is ready for application after optimization. Each path in the model can present a specific EMI event.
After that, the EMC problems of the target system are presented via a network based on the EN model.

Traction System and the OTCS in EMU
The OTCS equipment is one of the important technical equipment to ensure the safe operation of the EMU in high-speed railway. The EMC problem in OTCS is mainly caused by the traction system in EMU, so the traction system and OTCS are first analyzed to obtain the information and materials and how to model.

Traction System
The traction system of the EMU consisted of the pantograph, high-voltage cable, traction transformer, traction converter, auxiliary power unit, traction motors, and ground device as shown in Figure 5. From the physical and electrical structure, it can be known that most of the devices and components of the traction system are concentrated on the top and bottom space of the EMU, except the pantograph on the roof. The electric energy is obtained through sliding on the contact wire of the catenary by the pantograph, and the motors propel the EMU in the traction drive system. As the conclusions in References [4,8], pantograph arcing is one of the sources in EMU.
There are three main channels for EM disturbances from the outside of the train to enter the inside of it: (1) radiation coupling by antennas on the roof and bottom of the vehicle, (2) direct radiation coupling in the body space, (3) conduction coupling related to the channel of "pantograph-traction transformer-traction converter-power line-ground device". However, EM disturbances from the inside of the train are more complex, especially the auxiliary circuits (auxiliary converters, batteries, and battery chargers, etc.), air conditioning, traction converters, and traction motors make the coupling complicated.  There are three main channels for EM disturbances from the outside of the train to enter the inside of it: (1) radiation coupling by antennas on the roof and bottom of the vehicle, (2) direct radiation coupling in the body space, (3) conduction coupling related to the channel of "pantograph-traction transformer-traction converter-power line-ground device". However, EM disturbances from the inside of the train are more complex, especially the auxiliary circuits (auxiliary converters, batteries, and battery chargers, etc.), air conditioning, traction converters, and traction motors make the coupling complicated.

OTCS Equipment
The core function of the OTCS equipment is to supervise the speed of EMU in real-time and automatically control the braking system to prevent the train from over-speeding. A brief introduction of the CTCS-3 (Chinese Train Control System Level 3) level OTCS is given as an example below [23,24].
The subsystems and devices of the OTCS equipment cooperate with the wayside equipment to achieve real-time control. Figure 6 shows the structure diagram and locations of the OTCS, the main functions and features of the equipment are as follows: Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 18

OTCS Equipment
The core function of the OTCS equipment is to supervise the speed of EMU in real-time and automatically control the braking system to prevent the train from over-speeding. A brief introduction of the CTCS-3 (Chinese Train Control System Level 3) level OTCS is given as an example below [23,24].
The subsystems and devices of the OTCS equipment cooperate with the wayside equipment to achieve real-time control. Figure 6 shows the structure diagram and locations of the OTCS, the main functions and features of the equipment are as follows:

•
The roof of the EMU The location and line data in the movement authority (MA) are sent and received by the antenna on the roof of the EMU between the OTCS and the wayside equipment via wireless communication. GSM-R is the most widely used and LTE-R (Long-Term Evolution for Railway) will be applied in the future.

•
The bottom of the EMU The speed sensor is used to acquire the speed and the over-speed protection is realized by the speed and distance processing unit (SDU) in OTCS with the acceleration data from the radar. The parts of the MA information from the track circuit and balise are acquired by the track circuit receiver (TCR) and compact antenna unit (CAU) in the balise transmission module (BTM). BTM can also calculate and calibrate the location in real-time.

•
The insides of the EMU The driver-machine interface (DMI) is used to show the interaction information to the driver and the driver can execute some instructions through DMI to control the EMU. The train interface unit (TIU) is used to provide an interface between the train and OTCS. The kernel of the OTCS and BTM are installed in the ATP cabinet with power and ground connection with the EMU.

EN Model of the OTCS
Considering the symmetry of the EMU, it takes 4-car of the 8-car EMU to model the OTCS for simplicity. Because there are two OTCS and traction systems in an 8-car EMU for redundancy. According to the procedure in Section 2, the nodes were deleted and merged based on the statistical data, expert knowledge, and literature. Then, the 26 nodes of the EN model are extracted and shown in Table 3 based on the structure in Figures 5 and 6.

•
The roof of the EMU The location and line data in the movement authority (MA) are sent and received by the antenna on the roof of the EMU between the OTCS and the wayside equipment via wireless communication. GSM-R is the most widely used and LTE-R (Long-Term Evolution for Railway) will be applied in the future.

•
The bottom of the EMU The speed sensor is used to acquire the speed and the over-speed protection is realized by the speed and distance processing unit (SDU) in OTCS with the acceleration data from the radar. The parts of the MA information from the track circuit and balise are acquired by the track circuit receiver (TCR) and compact antenna unit (CAU) in the balise transmission module (BTM). BTM can also calculate and calibrate the location in real-time.

•
The insides of the EMU The driver-machine interface (DMI) is used to show the interaction information to the driver and the driver can execute some instructions through DMI to control the EMU. The train interface unit (TIU) is used to provide an interface between the train and OTCS. The kernel of the OTCS and BTM are installed in the ATP cabinet with power and ground connection with the EMU.

EN Model of the OTCS
Considering the symmetry of the EMU, it takes 4-car of the 8-car EMU to model the OTCS for simplicity. Because there are two OTCS and traction systems in an 8-car EMU for redundancy. According to the procedure in Section 2, the nodes were deleted and merged based on the statistical data, expert knowledge, and literature. Then, the 26 nodes of the EN model are extracted and shown in Table 3 based on the structure in Figures 5 and 6.

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1.

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and defi EN model. The model construction is divided into several parts, includin representation of the node, the definition and form of the edge, the m simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC el components in the target system. In other words, each node in the EN model, w one-to-one match with the EMC elements and system functional structure, has a meaning. Thus, the attributes or characteristic parameters of the nodes can calculation, simulation, or measurements, such as the amplitude, frequency, t the module function. Based on the above definition, the node form can be deter

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give where EN denotes the EN model; Av and Ae are the attribute function for the nod thus Av and Ae can be represented by many forms such as simulation model, t so on; V and E denote the sets of the nodes and edges; and N is the node numb 4. Procedure The overall procedure has the following five phases according to the preli

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give where EN denotes the EN model; Av and Ae are the attribute function for the nod thus Av and Ae can be represented by many forms such as simulation model, t so on; V and E denote the sets of the nodes and edges; and N is the node numb 4. Procedure The overall procedure has the following five phases according to the preli v 4 Traction system (high voltage cable) The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2. The edge between the source and coupling nodes are the same with coupling connection. 2 Field-line coupling and inductive or capacitive coupling (crosstalk) are classified as this form for simplicity.

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the results of the network achieve an acceptable level, the process is stopped and the model is analyzed in the next step. Otherwise, the model should be simplified by several times until there are no repetition and conflict between the nodes and edges. After the work above, the expression can be given with the format as: where EN denotes the EN model; Av and Ae are the attribute function for the node and edge separately, thus Av and Ae can be represented by many forms such as simulation model, transfer function, and so on; V and E denote the sets of the nodes and edges; and N is the node number in the EN model.

Procedure
The overall procedure has the following five phases according to the preliminary above: The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give where EN denotes the EN model; Av and Ae are the attribute function for the nod thus Av and Ae can be represented by many forms such as simulation model, t so on; V and E denote the sets of the nodes and edges; and N is the node numb 4. Procedure The overall procedure has the following five phases according to the preli

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows: 1. The definition and representation of the node The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give where EN denotes the EN model; Av and Ae are the attribute function for the nod thus Av and Ae can be represented by many forms such as simulation model, t so on; V and E denote the sets of the nodes and edges; and N is the node numb 4. Procedure The overall procedure has the following five phases according to the preli v 6 Internal disturbance (insides of the EMU)

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows: 1. The definition and representation of the node The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give where EN denotes the EN model; Av and Ae are the attribute function for the nod thus Av and Ae can be represented by many forms such as simulation model, t so on; V and E denote the sets of the nodes and edges; and N is the node numb 4. Procedure The overall procedure has the following five phases according to the preli v 7 Traction system (main circuit)

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give where EN denotes the EN model; Av and Ae are the attribute function for the nod thus Av and Ae can be represented by many forms such as simulation model, t so on; V and E denote the sets of the nodes and edges; and N is the node numb 4. Procedure

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give where EN denotes the EN model; Av and Ae are the attribute function for the nod thus Av and Ae can be represented by many forms such as simulation model, t so on; V and E denote the sets of the nodes and edges; and N is the node numb v 9 Auxiliary power (radiation)

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give where EN denotes the EN model; Av and Ae are the attribute function for the nod thus Av and Ae can be represented by many forms such as simulation model, t so on; V and E denote the sets of the nodes and edges; and N is the node numb v 10 External disturbance at bottom

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give There are two kinds of EM disturbance nodes in the running scenario of the high-speed railway. One is the radiated disturbance node and the other is the conducted disturbance node. The attributes of the source node can be represented by waveforms in standards or the parameters measured onsite. Besides, two virtual nodes are designed, which are the radiation and conduction coupling nodes. There are two advantages, on the one hand, the virtual node can be used to determine the coupling relationship between the disturbance source and the module nodes. On the other hand, it can reduce the complexity of the network by reducing the cross among the connection edges from the disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC network domain, the connection relationship between the disturbance source and equipment is divided into two types, namely, the conduction and radiation coupling connections correspond to the coupling nodes in Table 1. Additionally, to address the connection in the system structure network domain, such as signal connection and physical connection, the form of the edge are designed in Table 2.

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the results of the network achieve an acceptable level, the process is stopped and the model is analyzed in the next step. Otherwise, the model should be simplified by several times until there are no repetition and conflict between the nodes and edges. After the work above, the expression can be given with the format as: The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give Radiation coupling (Virtual node)

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give Radiation coupling (Virtual node) parts, the model is named the EN model for simplicity.

Approach of the Network Modeling
In this subsection, we aim to introduce the step-by-step approach and definitions applied in the EN model. The model construction is divided into several parts, including the definition and representation of the node, the definition and form of the edge, the model expression and simplification, and the procedure. Detailed descriptions are as follows:

The definition and representation of the node
The node is the basic unit of the network. It can represent the EMC elements and sensitive components in the target system. In other words, each node in the EN model, which has a consistent one-to-one match with the EMC elements and system functional structure, has a specific function and meaning. Thus, the attributes or characteristic parameters of the nodes can be obtained through calculation, simulation, or measurements, such as the amplitude, frequency, transfer function, and the module function. Based on the above definition, the node form can be determined in Table 1. The connections between disturbance and equipment.
There are two kinds of EM disturbance nodes in the running scenario of th One is the radiated disturbance node and the other is the conducted disturbanc of the source node can be represented by waveforms in standards or the para site. Besides, two virtual nodes are designed, which are the radiation and condu There are two advantages, on the one hand, the virtual node can be used to de relationship between the disturbance source and the module nodes. On the oth the complexity of the network by reducing the cross among the connec disturbance nodes to the module nodes.

The definition and form of the edge
Considering the coupling process of the EM disturbance in the EMC n connection relationship between the disturbance source and equipment is div namely, the conduction and radiation coupling connections correspond to th Table 1. Additionally, to address the connection in the system structure netw signal connection and physical connection, the form of the edge are designed in The edge between the source and coupling nodes are the same with coupling conn coupling and inductive or capacitive coupling (crosstalk) are classified as this form

The model simplification and expression
Based on the definition of the node and edge in the EN model, if the r achieve an acceptable level, the process is stopped and the model is analy Otherwise, the model should be simplified by several times until there are no r between the nodes and edges. After the work above, the expression can be give Based on Table 3, we accomplished the EN model of the OTCS, which is described in Figure 7. There are four areas in the model and most of the connections are concentrated on the top and bottom areas. For the EMI events research, specific data, simulation model, and measurement will be presented and analyzed for each relevant element of the nodes and edges in Figure 7. However, it seems to be a difficult task for acquiring all of the possible paths in the model once time. The EN model of the OTCS will be updated based on the reports and research activities in the authority organization if a new EMI event occurs.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 9 of 18 seems to be a difficult task for acquiring all of the possible paths in the model once time. The EN model of the OTCS will be updated based on the reports and research activities in the authority organization if a new EMI event occurs.

Case Study
Section 3 has constructed the EN model of the OTCS to have a system-level view of the EMC topic. If certain properties of the nodes and edges can be found to analyze the paths from the on-site reports, the model can be used to resolve the new EMI event with the updated paths.

Paths Description
For the application of the EN model introduced in Section 2 and 3, this section contains the actual

Case Study
Section 3 has constructed the EN model of the OTCS to have a system-level view of the EMC topic. If certain properties of the nodes and edges can be found to analyze the paths from the on-site reports, the model can be used to resolve the new EMI event with the updated paths.

Paths Description
For the application of the EN model introduced in Sections 2 and 3, this section contains the actual cases of specific EMI events involving the disturbance and coupling. The content of the cases is presented in the subsequent sections. Figure 8 illustrates the paths with red and green lines concerning the interaction paths between the source and equipment. The LTE-R (LTE for railway) communications technology will be applied in the future and it has experimented with test lines by the railway group in China. The applicability of LTE-R in the 450 MHz frequency band should be considered according to the existing spectrum resource distribution in China. The transient radiated electric field from the pantograph arcing will have a potential impact on the performance of LTE-R at 450 MHz operation frequency [25]. The path of this case is the red line in Figure 8.
DMI (Driver-machine Interface unit) displays the information of control order, equipment status, and driving strategies of the train, as well as the fault information to the driver. It is located in the driver room connecting the kernel in OTCS by long cables. So, it could be influenced by the interaction between cables especially the nearby power lines in EMU. The path of this case is the green line in Figure 8.
Therefore, to verify the proposed modeling method for EMC in OTCS, these paths are used as practical examples in this study.

Paths in the Model
Although the explanation of the nodes can be found in Table 3, the reason for the events is still not investigated in detail. Here, we try to give the basic information and the attributions of the nodes and edges. Figure 9 gives the path of the EMI event in case A. As we discussed above, v2 is the disturbance from the pantograph arcing and v14 is the virtual coupling node of radiation. v17 is the antenna of the LTE-R and v26 is the function part. Then, the three edges are e2-14, e14-17, and e17-26 correspondingly. In other words, for radiation coupling represented by v14, the pantograph arcing in v2 will be received at v17 and the influence on v26 can be reflected in an index of its function. It should be noted that not all of the attributions of the nodes and edges have specific values. Some of them indicate the deterministic value for existence. In path A, Ae2-14 Av14, and Ae17-26 are assigned with the deterministic value of 1. It indicates the path that happened in this scenario. On the contrary, Av2, Ae14-17, and Av26 will be represented by measurement and simulation. In this case, Av17 is the measurement results from the receiving antenna, and Av26 can be considered as a result output module with a specific index by simulation. The LTE-R (LTE for railway) communications technology will be applied in the future and it has experimented with test lines by the railway group in China. The applicability of LTE-R in the 450 MHz frequency band should be considered according to the existing spectrum resource distribution in China. The transient radiated electric field from the pantograph arcing will have a potential impact on the performance of LTE-R at 450 MHz operation frequency [25]. The path of this case is the red line in Figure 8.
DMI (Driver-machine Interface unit) displays the information of control order, equipment status, and driving strategies of the train, as well as the fault information to the driver. It is located in the driver room connecting the kernel in OTCS by long cables. So, it could be influenced by the interaction between cables especially the nearby power lines in EMU. The path of this case is the green line in Figure 8.
Therefore, to verify the proposed modeling method for EMC in OTCS, these paths are used as practical examples in this study.

Paths in the Model
Although the explanation of the nodes can be found in Table 3, the reason for the events is still not investigated in detail. Here, we try to give the basic information and the attributions of the nodes and edges. Figure 9 gives the path of the EMI event in case A. As we discussed above, v 2 is the disturbance from the pantograph arcing and v 14 is the virtual coupling node of radiation. v 17 is the antenna of the LTE-R and v 26 is the function part. Then, the three edges are e 2-14 , e [14][15][16][17] , and e 17-26 correspondingly. In other words, for radiation coupling represented by v 14 , the pantograph arcing in v 2 will be received at v 17 and the influence on v 26 can be reflected in an index of its function. It should be noted that not all of the attributions of the nodes and edges have specific values. Some of them indicate the deterministic value for existence. In path A, A e2-14 A v14 , and A e17-26 are assigned with the deterministic value of 1. It indicates the path that happened in this scenario. On the contrary, A v2 , A e14-17 , and A v26 will be represented by measurement and simulation. In this case, A v17 is the measurement results from the receiving antenna, and A v26 can be considered as a result output module with a specific index by simulation. and edges. Figure 9 gives the path of the EMI event in case A. As we discussed above, v2 is the disturbance from the pantograph arcing and v14 is the virtual coupling node of radiation. v17 is the antenna of the LTE-R and v26 is the function part. Then, the three edges are e2-14, e14-17, and e17-26 correspondingly. In other words, for radiation coupling represented by v14, the pantograph arcing in v2 will be received at v17 and the influence on v26 can be reflected in an index of its function. It should be noted that not all of the attributions of the nodes and edges have specific values. Some of them indicate the deterministic value for existence. In path A, Ae2-14 Av14, and Ae17-26 are assigned with the deterministic value of 1. It indicates the path that happened in this scenario. On the contrary, Av2, Ae14-17, and Av26 will be represented by measurement and simulation. In this case, Av17 is the measurement results from the receiving antenna, and Av26 can be considered as a result output module with a specific index by simulation. Figure 9. The path of case A. Figure 9. The path of case A. Figure 10 gives the path of the EMI event in case B. v 5 is the disturbance from power lines in the auxiliary power and v 13 is the virtual coupling node of cable coupling. v 16 is the cable between the DMI device and the kernel v 20 in OTCS. Then, the three edges are e 5-13 , e [13][14][15][16] , and e 16-20 correspondingly. The path has the following scene meaning, for cable coupling represented by v 13 , the disturbance in power line in v 5 will be received by v 16 and the influence on v 20 can be reflected on the device v 15 . If the signal cannot be received within a certain time, the DMI v 15 will display a "communication interruption error". At this time, the v 20 would give the emergency brake control order to the EMU leading to temporary parking. In this path, A e5-13 , A e16-20 , and A v13 are assigned with the deterministic value of 1. It indicates the path that happened in this scenario. A v5 , A v16 , and A e13-16 will be represented by simulation and measurement. A v20 is the signal results at the cable terminal.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 11 of 18 Figure 10 gives the path of the EMI event in case B. v5 is the disturbance from power lines in the auxiliary power and v13 is the virtual coupling node of cable coupling. v16 is the cable between the DMI device and the kernel v20 in OTCS. Then, the three edges are e5-13, e13-16, and e16-20 correspondingly. The path has the following scene meaning, for cable coupling represented by v13, the disturbance in power line in v5 will be received by v16 and the influence on v20 can be reflected on the device v15. If the signal cannot be received within a certain time, the DMI v15 will display a "communication interruption error". At this time, the v20 would give the emergency brake control order to the EMU leading to temporary parking. In this path, Ae5-13, Ae16-20, and Av13 are assigned with the deterministic value of 1. It indicates the path that happened in this scenario. Av5, Av16, and Ae13-16 will be represented by simulation and measurement. Av20 is the signal results at the cable terminal.

Analysis and Discussion
The pantograph arcing is selected as the attribute of Av2. To obtain the actual waveform of the pantograph arcing during the operation of the EMU, a measuring probe is installed at the position of the train roof (see Figure 11a). As shown in the figure, the horizontal distance between the pantograph and the measurement antenna is 2.5 m and the antenna is 0.4 m in height above the roof. A coaxial cable is fixed on the outside of the train for connecting the instrument temporarily. The receiver and oscilloscope are set inside of the train under the control of the computer for data collection.  The pantograph arcing is selected as the attribute of A v2 . To obtain the actual waveform of the pantograph arcing during the operation of the EMU, a measuring probe is installed at the position of the train roof (see Figure 11a). As shown in the figure, the horizontal distance between the pantograph and the measurement antenna is 2.5 m and the antenna is 0.4 m in height above the roof. A coaxial cable is fixed on the outside of the train for connecting the instrument temporarily. The receiver and oscilloscope are set inside of the train under the control of the computer for data collection.
The single waveform of the EM disturbance from the pantograph arcing that is measured on-site is shown in Figure 11b. The duration of the single discharge is within 1 µs and the rise time is at nanoseconds (ns) level. However, the length of a symbol is around tens of µs in the communication system, i.e., the length of a symbol is about 72 µs in FDD-LTE (Long-Term Evolution Frequency-Division Duplex) with OFDM (Orthogonal Frequency-Division Multiplexing). Thus, the single waveform cannot cover the useful signals completely. It should be noted that most of the pantograph arcing from the off-line discharge is not a single pulse signal. There will be multiple arcing phenomena while the off-line between pantograph and catenary occur by sliding contact. The reason is that the off-line is continuous until the displacement difference between them is no longer zero. Therefore, it is assumed that the pantograph arcing disturbance is composed of multiple independent pulse signals like the waveform in Figure 11b. To get the time interval (TI) of the pulse, statistical analysis of the TI in 50 sets of scanning results is done based on the Weibull distribution. Thus, the continuous disturbance composed of the waveform in Figure 11b with about 0.17 ms TI is regarded as the attribute of v 2 .

Av2 of the node v2
The pantograph arcing is selected as the attribute of Av2. To obtain the actual waveform of the pantograph arcing during the operation of the EMU, a measuring probe is installed at the position of the train roof (see Figure 11a). As shown in the figure, the horizontal distance between the pantograph and the measurement antenna is 2.5 m and the antenna is 0.4 m in height above the roof. A coaxial cable is fixed on the outside of the train for connecting the instrument temporarily. The receiver and oscilloscope are set inside of the train under the control of the computer for data collection. The single waveform of the EM disturbance from the pantograph arcing that is measured onsite is shown in Figure 11b. The duration of the single discharge is within 1 μs and the rise time is at nanoseconds (ns) level. However, the length of a symbol is around tens of μs in the communication system, i.e., the length of a symbol is about 72 μs in FDD-LTE (Long-Term Evolution Frequency-Division Duplex) with OFDM (Orthogonal Frequency-Division Multiplexing). Thus, the single waveform cannot cover the useful signals completely. It should be noted that most of the pantograph arcing from the off-line discharge is not a single pulse signal. There will be multiple arcing phenomena while the off-line between pantograph and catenary occur by sliding contact. The reason

2.
A e14-17 of the edge e [14][15][16][17] In general, there are two kinds of radiated parts from the structure including pantograph, contact wire, and train roof while the arc discharge happens. The primary direct radiation is from the arc itself and the other is the secondary radiation generated by the induced current in the pantograph and contact wire. Therefore, the signal in v 2 is the excitation of the structure of the pantograph-contact wire and the energy will be radiated by the equivalent antenna. Considering the large dimension in the longitudinal direction of the contact wire in a catenary, the equivalent antenna can be regarded as the Traveling Wave Wire Antenna. Thus, A e14-17 of the edge e 14-17 can be represented by the model of the whole structure including pantograph, contact wire, and train roof in CST software as shown in Figure 12.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 12 of 18 is that the off-line is continuous until the displacement difference between them is no longer zero. Therefore, it is assumed that the pantograph arcing disturbance is composed of multiple independent pulse signals like the waveform in Figure 11b. To get the time interval (TI) of the pulse, statistical analysis of the TI in 50 sets of scanning results is done based on the Weibull distribution. Thus, the continuous disturbance composed of the waveform in Figure 11b with about 0.17 ms TI is regarded as the attribute of v2.

Ae14-17 of the edge e14-17
In general, there are two kinds of radiated parts from the structure including pantograph, contact wire, and train roof while the arc discharge happens. The primary direct radiation is from the arc itself and the other is the secondary radiation generated by the induced current in the pantograph and contact wire. Therefore, the signal in v2 is the excitation of the structure of the pantograph-contact wire and the energy will be radiated by the equivalent antenna. Considering the large dimension in the longitudinal direction of the contact wire in a catenary, the equivalent antenna can be regarded as the Traveling Wave Wire Antenna. Thus, Ae14-17 of the edge e14-17 can be represented by the model of the whole structure including pantograph, contact wire, and train roof in CST software as shown in Figure 12. In the model in Figure 12, the excitation is modeled as a discrete source between contact wire and pantograph with the continuous disturbance from v2, and the gap between pantograph and contact line is built as 5 mm. v17 is the receiver at the end of the roof. In this model, the size of the train roof is set at 25 m × 4 m × 0.03 m with the material aluminum. Since the roof is the main reflective structure, the carriage is simplified and represented by the roof. The height of the contact line and the roof is set to 2 m. The vertical distance between the contact line and the roof is set at 2 m. Since the actual length of the contact line is much longer than the train, the load is added on the contact line terminals at a quarter-wavelength position to avoid the influence in the simulation.
Based on the above analysis, the node of v17 is simulated as a receiving probe corresponding to the function in OTCS. Thus, the path v2 → v14 → v17 that the propagation characteristics could be analyzed according to the model above. Furthermore, the electric field at the v17 position could be obtained by this model. However, it is different from the real environment in the high-speed railway. To get the actual signal for the next segment in the path, the signal is measured at the node of v17 and the result is shown in Figure 13. The pulse received at the v17 position is in Figure 13a and it has the same trends in Figure 11b. The port voltage received at the antenna cable of the module node v26 is In the model in Figure 12, the excitation is modeled as a discrete source between contact wire and pantograph with the continuous disturbance from v 2 , and the gap between pantograph and contact line is built as 5 mm. v 17 is the receiver at the end of the roof. In this model, the size of the train roof is set at 25 m × 4 m × 0.03 m with the material aluminum. Since the roof is the main reflective structure, the carriage is simplified and represented by the roof. The height of the contact line and the roof is set to 2 m. The vertical distance between the contact line and the roof is set at 2 m. Since the actual length of the contact line is much longer than the train, the load is added on the contact line terminals at a quarter-wavelength position to avoid the influence in the simulation.
Based on the above analysis, the node of v 17 is simulated as a receiving probe corresponding to the function in OTCS. Thus, the path v 2 → v 14 → v 17 that the propagation characteristics could be analyzed according to the model above. Furthermore, the electric field at the v 17 position could be obtained by this model. However, it is different from the real environment in the high-speed railway. To get the actual signal for the next segment in the path, the signal is measured at the node of v 17 and the result is shown in Figure 13. The pulse received at the v 17 position is in Figure 13a and it has the same trends in Figure 11b. The port voltage received at the antenna cable of the module node v 26 is shown in Figure 13b. Thus, the analysis process indicates the feasibility of the path v 2 → v 14 → v 17 in the EN model.

Av26 of the node v26
To complete the case analysis, a downlink of LTE-R is established in Matlab for evaluating the path e17-26 as shown in Figure 14. Since the throughput is an important criterion for evaluating the communication system, the throughput, as a basic index, is chosen as the attribute Av26 of node v26. Thus, the path v2 → v14 → v17 → v26 could be complete after the study above.
In the downlink model of e17-26, PDSCH (Physical Downlink Shared Channel) with modulation of 64QAM is selected for downlink simulation by the single antenna transceiver and the carrier frequency is 450 MHz with FDD standard [26]. The channel is AWGN and the bandwidth is 5 MHz. The port voltage received after the node v17 at the cable end in Figure 13b is added to the transmission channel. The relationship between the throughput and SNR (Signal to noise ratio) is shown in Figure  15 with the blue line. The throughput decreases when the disturbance caused by the pantograph arcing from the node v2.

3.
A v26 of the node v 26 To complete the case analysis, a downlink of LTE-R is established in Matlab for evaluating the path e 17-26 as shown in Figure 14. Since the throughput is an important criterion for evaluating the communication system, the throughput, as a basic index, is chosen as the attribute A v26 of node v 26 . Thus, the path v 2 → v 14 → v 17 → v 26 could be complete after the study above.

Av26 of the node v26
To complete the case analysis, a downlink of LTE-R is established in Matlab for evaluating the path e17-26 as shown in Figure 14. Since the throughput is an important criterion for evaluating the communication system, the throughput, as a basic index, is chosen as the attribute Av26 of node v26. Thus, the path v2 → v14 → v17 → v26 could be complete after the study above.
In the downlink model of e17-26, PDSCH (Physical Downlink Shared Channel) with modulation of 64QAM is selected for downlink simulation by the single antenna transceiver and the carrier frequency is 450 MHz with FDD standard [26]. The channel is AWGN and the bandwidth is 5 MHz. The port voltage received after the node v17 at the cable end in Figure 13b is added to the transmission channel. The relationship between the throughput and SNR (Signal to noise ratio) is shown in Figure  15 with the blue line. The throughput decreases when the disturbance caused by the pantograph arcing from the node v2.  In the downlink model of e 17-26 , PDSCH (Physical Downlink Shared Channel) with modulation of 64QAM is selected for downlink simulation by the single antenna transceiver and the carrier frequency is 450 MHz with FDD standard [26]. The channel is AWGN and the bandwidth is 5 MHz. The port voltage received after the node v 17 at the cable end in Figure 13b is added to the transmission channel. The relationship between the throughput and SNR (Signal to noise ratio) is shown in Figure 15 with the blue line. The throughput decreases when the disturbance caused by the pantograph arcing from the node v 2 .
Appl. Sci. 2020, 10, x FOR PEER REVIEW 14 of 18 Figure 15. The evaluation criterion for Av26 of the node v26 with the relationship throughput and SNR.
In Figure 15, the measurement result is from the work in Reference [25]. It indicates the result is consistent with the simulation in e17-26. It is based on the FTP (File Transfer Protocol) to measure the throughput of the LTE-R system in an experimental high-speed railway line in Shenyang, China. The on-board and wayside terminal will transfer a 100 MB file when the train passes through the coverage area of the network. Then, the device calculates the maximum number of bytes per second and we fit the relationship between the downlink throughput and the SNR.
The measurement results verify the validity of the simulation of the path v17 → v26. Thus, the EMI event of v2 → v14 → v17 → v26 in the EN model of the OTCS has been analyzed and the results show that the pantograph arcing disturbance will affect the performance of communication.

Path B
1. Av5 of the node v5 As the description in Section 3.1.1, the conductive EM disturbances from the channel of "pantograph-traction transformer-converter-power line-ground device" will make the interactions between the power system and OTCS equipment. The factory workers may lay the cable between DMI device v15 and the OTCS kernel v20 at the installation space together with the power cables inside the train body for saving space. However, this condition leads to the EMI event in path B occurs.
The waveform in Figure 16 is selected as the attribute of Av5 as the equivalent conduction disturbance from the railway system by measurement [27]. The amplitude voltage of the actual measurement transient signal is about 5.3 V. Thus, the disturbance showed in Figure 16 with 2.5 μs duration is regarded as one signal in Av5. In Figure 15, the measurement result is from the work in Reference [25]. It indicates the result is consistent with the simulation in e [17][18][19][20][21][22][23][24][25][26] . It is based on the FTP (File Transfer Protocol) to measure the throughput of the LTE-R system in an experimental high-speed railway line in Shenyang, China. The on-board and wayside terminal will transfer a 100 MB file when the train passes through the coverage area of the network. Then, the device calculates the maximum number of bytes per second and we fit the relationship between the downlink throughput and the SNR.
The measurement results verify the validity of the simulation of the path v 17 → v 26 . Thus, the EMI event of v 2 → v 14 → v 17 → v 26 in the EN model of the OTCS has been analyzed and the results show that the pantograph arcing disturbance will affect the performance of communication.

1.
A v5 of the node v 5 As the description in Section 3.1.1, the conductive EM disturbances from the channel of "pantograph-traction transformer-converter-power line-ground device" will make the interactions between the power system and OTCS equipment. The factory workers may lay the cable between DMI device v 15 and the OTCS kernel v 20 at the installation space together with the power cables inside the train body for saving space. However, this condition leads to the EMI event in path B occurs.
The waveform in Figure 16 is selected as the attribute of A v5 as the equivalent conduction disturbance from the railway system by measurement [27]. The amplitude voltage of the actual measurement transient signal is about 5.3 V. Thus, the disturbance showed in Figure 16 with 2.5 µs duration is regarded as one signal in A v5 . cables inside the train body for saving space. However, this condition leads to the EMI event in path B occurs.
The waveform in Figure 16 is selected as the attribute of Av5 as the equivalent conduction disturbance from the railway system by measurement [27]. The amplitude voltage of the actual measurement transient signal is about 5.3 V. Thus, the disturbance showed in Figure 16 with 2.5 μs duration is regarded as one signal in Av5. Figure 16. The waveform from the node v5. Figure 16. The waveform from the node v 5 .

2.
A e13-16 of the edge e [13][14][15][16] According to the layout of the EMU, the power wire v 5 is a long cable with the 220 V, 50 Hz AC, and the resistance at both ends is 50 Ω. The DMI cable v 16 is placed parallel to v 5 . Therefore, the signal in v 5 is the source of the EMI path and the terminal at v 20 will receive the disturbance signal. Considering the basic shape of the EMU, a rectangular body is used to simulate the train body. The size of the train is set at 25 m × 4 m × 4 m with 0.03 mm material aluminum in thickness. The DMI cable is close to the sidewall of the train body, 30 cm away from the ground and the sidewall. The power line and the DMI cable are routed in parallel, and the length is 20 m, as shown in Figure 17. Then, the core of the power cable is 0.75 mm 2 , and the core of the DMI cable is 0.5 mm 2 , and the shielding thickness of them is both 0.5 mm. The terminal v 20 is connected to the 120 Ω load, and the terminal v 5 is connected to a load of 150 Ω. According to the layout of the EMU, the power wire v5 is a long cable with the 220 V, 50 Hz AC, and the resistance at both ends is 50 Ω. The DMI cable v16 is placed parallel to v5. Therefore, the signal in v5 is the source of the EMI path and the terminal at v20 will receive the disturbance signal. Considering the basic shape of the EMU, a rectangular body is used to simulate the train body. The size of the train is set at 25 m × 4 m × 4 m with 0.03 mm material aluminum in thickness. The DMI cable is close to the sidewall of the train body, 30 cm away from the ground and the sidewall. The power line and the DMI cable are routed in parallel, and the length is 20 m, as shown in Figure 17. Then, the core of the power cable is 0.75 mm 2 , and the core of the DMI cable is 0.5 mm 2 , and the shielding thickness of them is both 0.5 mm. The terminal v20 is connected to the 120 Ω load, and the terminal v5 is connected to a load of 150 Ω. Based on the above analysis and configuration, the node v20 is simulated as the port signal at end of the cable v16 corresponding to the receiver function in OTCS. Then, the path v5 → v16 → v20 that the cable coupling EMI event could be analyzed according to the model above.

Av20 of the node v20
In the first scene, it is assumed that the node v15 sends a square wave signal with the amplitude of 2.5 V, the duration is 0.07 ms, and the rising time is 1 μs. The result in Figure 18a is the ideal waveform, that is, there is no power cable nearby, the cable v16 is only influenced by the body and grounding current. The waveform in Figure18b is the second scene where the power cable v5 is placed next to the cable v16. In Figure 18, it can be known that the cable coupling has affected the node v16, however, the communication function between nodes v15 and v20 is still normal due to the low received disturbance level. Based on the above analysis and configuration, the node v 20 is simulated as the port signal at end of the cable v 16 corresponding to the receiver function in OTCS. Then, the path v 5 → v 16 → v 20 that the cable coupling EMI event could be analyzed according to the model above.

3.
A v20 of the node v 20 In the first scene, it is assumed that the node v 15 sends a square wave signal with the amplitude of 2.5 V, the duration is 0.07 ms, and the rising time is 1 µs. The result in Figure 18a is the ideal waveform, that is, there is no power cable nearby, the cable v 16 is only influenced by the body and grounding current. The waveform in Figure 18b is the second scene where the power cable v 5 is placed next to the cable v 16 . In Figure 18, it can be known that the cable coupling has affected the node v 16 , however, the communication function between nodes v 15 and v 20 is still normal due to the low received disturbance level.
3. Av20 of the node v20 In the first scene, it is assumed that the node v15 sends a square wave signal with the amplitude of 2.5 V, the duration is 0.07 ms, and the rising time is 1 μs. The result in Figure 18a is the ideal waveform, that is, there is no power cable nearby, the cable v16 is only influenced by the body and grounding current. The waveform in Figure18b is the second scene where the power cable v5 is placed next to the cable v16. In Figure 18, it can be known that the cable coupling has affected the node v16, however, the communication function between nodes v15 and v20 is still normal due to the low received disturbance level.
(a) (b) Figure 18. The results at node v20. (a) No power cable near the DMI cable; (b) The signal received at the node v20 when the power cable is nearby. In the second scene, the disturbance in Figure 17 is superimposed on the power cable with a 220 V AC signal, and the result at node v 20 is shown in Figure 19a. The figure shows that the power cable has an impact on the signal of the DMI cable. Assuming that the transmission is the differential mode of RS485, thus, the difference between the two cores determines the terminal signal. It can be found in the figure that the voltage difference between cores has reached about 7 V, which obviously exceeds its normal decision level. Thus, it may cause the fault detailed in "communication interruption error".
Appl. Sci. 2020, 10, x FOR PEER REVIEW 16 of 18 In the second scene, the disturbance in Figure 17 is superimposed on the power cable with a 220 V AC signal, and the result at node v20 is shown in Figure 19a. The figure shows that the power cable has an impact on the signal of the DMI cable. Assuming that the transmission is the differential mode of RS485, thus, the difference between the two cores determines the terminal signal. It can be found in the figure that the voltage difference between cores has reached about 7 V, which obviously exceeds its normal decision level. Thus, it may cause the fault detailed in "communication interruption error".
(a) (b) Figure 19. The results at node v20 with the disturbance. (a) The power cable with the disturbance source; (b) The signal measured at the port of node v20 in OTCS.
To verify the analysis results, the measure results in time-domain at the port of node v20 in OTCS is shown in Figure 19b. Taking into account the actual cable length, the amplitude of the disturbance and the protective measures, the waveform of the measured result is relatively smooth than the simulation. However, the amplitude received has reached the same level in Figure 19a, which exceeds the normal judgment amplitude.
The simulation and measurement results verify the validity of the path v5 → v20. Thus, the EMI event of v5 → v13 → v16 → v20 in the EN model of the OTCS has been analyzed and the results show that the disturbance from the traction system would affect the communication between v15 and v20 if the power cable is nearby.

Conclusions
This paper proposes a network-based modeling method for EMI event analysis in OTCS, which decomposes the OTCS and EMU into nodes and edges to abstractly present the elements of EMC and equipment, making EMC analysis in system-level view possible. By combining the EM sources, sensitive equipment, and coupling into different paths in the EN model, each EMI event can be To verify the analysis results, the measure results in time-domain at the port of node v 20 in OTCS is shown in Figure 19b. Taking into account the actual cable length, the amplitude of the disturbance and the protective measures, the waveform of the measured result is relatively smooth than the simulation. However, the amplitude received has reached the same level in Figure 19a, which exceeds the normal judgment amplitude.
The simulation and measurement results verify the validity of the path v 5 → v 20 . Thus, the EMI event of v 5 → v 13 → v 16 → v 20 in the EN model of the OTCS has been analyzed and the results show that the disturbance from the traction system would affect the communication between v 15 and v 20 if the power cable is nearby.

Conclusions
This paper proposes a network-based modeling method for EMI event analysis in OTCS, which decomposes the OTCS and EMU into nodes and edges to abstractly present the elements of EMC and equipment, making EMC analysis in system-level view possible. By combining the EM sources, sensitive equipment, and coupling into different paths in the EN model, each EMI event can be studied from the origin node to the destination based on calculation, simulation, or measurement, respectively.
The modeling method based on network theory for EMI events under the high-speed railway EM environment is applied to the OTCS equipment in the EMU. The model has 26 nodes in this version with 9 disturbance nodes and 12 equipment nodes, respectively. The cases from the EMI event paths have been analyzed under the framework of the EN model, and the results indicate the new method is operable. Considering the analysis process, it is helpful for us to investigate EMI events and improve EMC performance in railway engineering. The modeling method describes the analysis process of the paths clearly and indicates the feasibility of the EN model. Besides, if all paths can be done like the cases and the new EMI event will be added in the updated model, a more integral model will be carried out to develop the EMC performance of the OTCS at the system-level in the future.