Testing Algorithms for Controlling the Distributed Power Supply System of a Railway Signal Box

Abstract

Trends in the use of renewable energy sources to power buildings do not bypass objects for which maintaining a power supply is critical. This also applies to railway signal boxes. The aim of the research work was to test the multisource power supply system for a railway signal box with power electronic converter systems and a DC bus, built as part of the research project. The assumption for powering the railway signal box building was to use renewable sources, energy storage devices, and a 3 kV DC traction network as the second required power supply grid. Both power grids were connected by power electronic converters, and the power values of the converters were set based on the calculated power balance values using the values measured at the system nodes and the set constraints. The tests primarily tested the response of the power supply system to changes in load power and power generated by the photovoltaic source, as well as the charge level of the energy storage devices. The correctness of the control algorithm’s operation was assessed based on the recorded power values in the power supply system nodes. The tests were carried out for 60 scenarios that covered all normal and emergency operating conditions. During the tests, delays in response to changes in the power supplied to the converters and the values of circular power flow between the power grid connections were recorded. The recorded delays ranged from 2 to about 50 s and the circular power flows did not exceed 1500 W. Based on the results of the tests, it was found necessary to improve the power measurement system in the power supply system nodes and to improve the quality of communication and the transmission time of measurement data transmission time.

1. Introduction

In recent years, with the increasing demand for energy and growing environmental awareness, effective energy management has become a priority in both the public and private sectors. The key challenge is not only to ensure the continuity of energy supply but also to minimize its costs and optimize the use of renewable energy sources. Energy management plays a crucial role in the context of the European Green Deal strategy [1] and the new Directive (EU) 2023/1791 [2]. This directive introduces measures to enhance energy efficiency in the European Union, including a legally binding target to reduce final energy consumption by 11.7% by 2030 compared to the 2020 baseline scenario.

An important element of the new directive is the “energy efficiency first principle”, which obliges EU member states to prioritize energy efficiency in their policies, planning, and investments related to the railway sector. In railway systems, especially in signal boxes, precise forecasting of energy demand and optimization of its consumption are becoming essential. These systems must be reliable and efficient, requiring advanced energy management methods and modern technologies [3].

Additionally, integrating energy management systems in buildings and other infrastructures, as reviewed by Al-Ghaili et al. [4], and the innovative strategies discussed by Shahee et al. [5], highlight the broader applicability of these principles.

Railway signal boxes require a dedicated power supply due to their crucial role in ensuring safety. They are powered by at least two independent power grids, 3 × 400 V AC (low-voltage grids), and a diesel generator. Additionally, key signaling elements have emergency battery power, which often requires the use of converters to improve energy quality, including reactive power compensation [6,7]. In Poland, the rules regarding the power supply of railway infrastructure are specified in Annex No. 3 to the resolution of 30 December 2019 [8]. Emergency power supplies must provide at least 2 h of power to all devices related to railway traffic. The railway signal box power supply system, in addition to the UPS battery systems, must be powered by at least two independent power grids and must also be equipped with a diesel generator. The integration of different energy sources, such as low-voltage power grids, traction grids, and renewable sources, poses a significant technological challenge [9,10].

In railway signal box buildings, the priority is to ensure power supply in any emergency situation. Access to renewable energy sources and energy storage technologies, as well as power converter systems, makes it possible to replace expensive connections to the power grid with local solutions. Due to the fact that the power supply system concerns railway signal boxes located near the traction grid, it was proposed to use this grid as a reserve source. Additionally, a power supply from a photovoltaic source was introduced. Energy storage is implemented in a lithium-ion battery and hydrogen. An innovative approach is the introduction of a DC bus to which all energy sources and storage facilities are connected. This solution reduces electrical energy losses during conversion [11,12].

The scientific literature includes numerous studies on energy management in railway systems and the integration of various energy sources. For example, Andruszkiewicz et al. [13] described the optimization of reactive power compensation levels in distribution grids, which has direct implications for the power supply of railway signal boxes. In studies on power converters, Sekar and Suresh [14] discussed suitable converters for renewable energy sources, which is crucial for integrating these sources into railway systems. Moreover, the use of multisource power supply systems is becoming increasingly common in various applications, including railway signal boxes, to enhance reliability and efficiency. Such systems are designed not only to ensure a continuous power supply but also to optimize the use of renewable energy and reduce operational costs [15,16]. Proper management of energy consumption is essential at all levels, from individual households to integrated energy systems. This hierarchy is justified because the growing popularity of distributed energy sources, especially renewable ones, forces the management and forecasting of the electricity demand to be transferred to lower levels [17,18]. In the case of railway systems, this applies, in particular, to signal boxes. Another significant challenge is to generate reliable demand and production forecasts, which are essential for precise planning and coordination, especially for systems utilizing renewable energy sources [19,20]. These forecasts enable more efficient and harmonized operation of the various components of the system. Consequently, this leads to the optimal use of energy resources and ensures the stability and reliability of energy supply in railway systems [21,22].

However, despite the numerous studies on energy management algorithms, there is a lack of comprehensive testing of these algorithms under real operational conditions. Detailed comparative analyses of different solutions and tests in various operational scenarios are needed. For example, Mystakidis et al. [23] provided a comprehensive review of energy forecasting techniques, highlighting the need for precise prediction methods in energy management systems. Furthermore, Tajjour and Chandel [24] conducted an experimental investigation of a novel smart energy management system for solar photovoltaic microgrids, demonstrating the importance of evaluating performance in real-time under varying conditions to improve reliability and efficiency. Additionally, studies by Zubair et al. [25] and Gümüş [26], Shi et al. [27], and Aman et al. [28] discuss the integration of renewable energy sources and advanced energy management techniques. These studies emphasize the critical need for extensive testing and comparison of energy management strategies to ensure their effectiveness in real-world applications.

Despite significant advances in energy management, there are clear research gaps in the testing of algorithms under real operational conditions of railway signal boxes. Current studies often do not cover the full range of operational scenarios, making it difficult to assess the practical effectiveness of proposed solutions. Furthermore, there is a lack of detailed comparisons with other solutions, making it difficult to determine the advantages and limitations of each approach. Particularly in the case of railway signal boxes with a critical power supply system.

The correctness of the algorithm means obtaining expected power values in all system power nodes for all tested scenarios, including the power imbalance in the system and the electrical power flows between grid connections. In [29], multiconverter solutions for connecting multiple power grids for a common DC bus and AC devices are shown. However, the solution shown concerns unidirectional converters. In such a system, there is no possibility of circular power flows between power grids. For the constructed power supply system, we assumed that the average power of the circular flow between the grid connections over a period of 10 s cannot exceed 1500 W.

The objective of this study is to test energy management algorithms in railway signal box power supply systems under various operational scenarios. This study aims to develop and test a new algorithm that integrates various energy sources, such as the traction grid, photovoltaics, lithium-ion batteries, and hydrogen storage. The article will present the results of tests conducted under laboratory conditions, considering different energy demand and generation scenarios. A very effective way to simulate control systems is to simulate and test the system in hardware in the loop mode [30]. The element of the control system control algorithms can be simulated in specialized programming environments, and parts of the power supply system are real systems. Unfortunately, in our case, it was not possible and we conducted system tests in conditions close to real conditions.

2. Materials and Methods

As part of the project, an algorithm for controlling the energy supply system has been developed. The task of the algorithm is mainly to provide power to the railway signal box building, especially for automated railway traffic control automation. Due to the completely different multisource power supply system, it was necessary to develop a new algorithm to control individual elements and manage electrical power in all power nodes of the system. The power supply system, in addition to local sources and energy storage, connects two external power lines: the national power grid 3 × 400 V AC and the traction grid 3 kV DC. In a typical solution, the railway signal box building is powered by two independent connections to 3 × 400 V AC power grids and switching between grids is carried out by contactors. It is not possible to connect two independent connections at the same time. In the constructed power supply system, two external power grids are connected to each other using power electronic converters, and the direction and value of power are controlled by the equations of the instantaneous power balance in the system. It was assumed that circular flows between connections cannot be allowed. The power nodes of the power balance in the power supply system depend on the equations in the algorithm and on the quality of the power measurements in the nodes and communication delays.

The algorithms tested for the railway signal box power supply system were developed with the objective of ensuring the security of the electrical supply in both normal and emergency operating conditions. The following sections present the construction of the railway signal box power supply system and the measurement laboratory. The basic assumptions for the control algorithm, as well as parts of the algorithm and the testing program, are also described.

2.1. Test Object and Measurement Setup

The test object was the power supply system of the railway signal box built as part of the research project. This system is characterized by the use of multiple power sources. Figure 1 shows a simplified diagram of the constructed power supply system, which includes the following sources of electricity: L1—3 × 400 V AC grid, L2—3 kV traction grid, photovoltaic source, and diesel generator. The system includes two energy storage units: a Li-ion battery and a hydrogen tank with fuel cell and electrolyzer. The signal box supply system also includes four main power electronic converters. A characteristic feature of the power supply system is the common DC bus and a battery directly connected to the DC bus. This allowed for a reduction in losses in an additional battery power converter. In addition, the traction grid and photovoltaic power converters do not have AC stages.

The electric system designed and manufactured for the railway signal box provides electricity supply in both on-grid and off-grid modes. In on-grid mode, it can perform selected target functions:

• Maximization of the use of energy from renewable sources;
• Reducing peak power consumption and output to the power grid;
• Reducing energy purchase costs.

In on-grid mode, power converters PL1 and PL2 are controlled by power value; the 630 V DC bus voltage is stabilized by the battery and simultaneously serves as a state-of-charge (SOC) indicator for the battery. The photovoltaic converter Ppv can operate with an MPPT algorithm, and in the case of a fully charged battery, it operates in maximum power limitation mode. The PH2 converter is controlled by the power value and the direction of energy flow, depending on the activated function: electrolyzer or fuel cell.

In the event of a power failure of the L1 3 × 400 V AC grid, the PL1 converter switches to off-grid mode where the voltage and frequency of the island grid are stabilized. The other converters continue to perform their previous functions.

This paper presents the results of tests carried out on the main algorithms that control the main functions of the railway signal box power supply system. To test the constructed power supply system, all components were installed on a test stand. The test stand is shown in Figure 2. Due to the lack of access to the traction grid, a 3 kV DC bidirectional grid simulator was built. The traction grid simulator includes a step-up and isolating transformer, a power converter that forms the L2 3 kV DC traction grid, and a chopper converter that dissipates excess energy during energy generation to the traction grid.

Figure 3 shows the test bench of the railway signal box power supply system. From a safety point of view, some parts of the 3 kV DC grid simulator are closed. The grid converter PL1 and the battery are located in a separate room, as it was necessary to provide a fireproof enclosure for the battery. All control and measurement components (except the 3 kV DC voltage measurement) are located in the control and measurement cabinet. The power of the converter unit is PL1—50 kW; PL2—50 kW; PH2—10 kW; and Ppv—10 kW.

To become independent of weather conditions and to set known and established operating conditions, two additional simulators were used:

• Photovoltaic source simulator (PV simulator): two series-connected power supplies Elektro-Automatik EA-PS 9200-40 (EA ELEKTRO-AUTOMATIK GMBH & CO. KG, Viersen, Germany) and ITECH IT6012C (ITECH ELECTRONIC CO., LTD., New Taipei, Taiwan). Due to the lack of a suitable programmable power supply, it was necessary to connect two power supplies in series to obtain a PV source voltage above 400 V DC. At this voltage, the Ppv power converter starts. The EA power supply worked as a constant voltage source, while the ITECH power supply had programmed characteristics of the PV cell. In this way, the Ppv converter could work with the MPPT algorithm. The Ppv power converter and PV simulator are shown in Figure 4a.
• Fuel cell/electrolyzer simulator (FC/EC simulator): bidirectional programmable power supply ITECH IT6012C-300-150 (ITECH ELECTRONIC CO., LTD., New Taipei, Taiwan) is shown in Figure 4b.

One of the design and construction assumptions for the power supply system is the ability to adapt the power supply configuration to the needs of the facility by selecting the necessary blocks. Another assumption is to ensure access to energy in case of failure of the power grid and other sources. For these reasons, the PL1 control system of the power grid inverter is directly communicated with the battery storage. The PL1 inverter controller is equipped with a CAN bus, the most popular in battery management systems (BMS). This ensures safe operation of the battery, as it is controlled directly by the PL1 controller, which is present in every power system configuration. Figure 5a shows the installed PL1 power converter and the battery module.

Battery parameters:

• Type—LiFePo4;
• Capacity—138 kWh;
• Rated voltage: 691 V;
• Output voltage range: 560 V to 790 V.

The 3 kV DC traction grid inverter was built as an independent module that can be installed outside the railway signal box building. This is due to the technical requirements and the safety of the premises. Figure 5b shows the PL2 converter block along with the 3 kV DC traction grid simulator.

The WAGO 762-5303 (WAGO GmbH & Co. KG, Minden, Germany) programmable controller is responsible for system control and communication with components of the supply system. The master controller communicates with the power converters via the RS-485 bus using the MODBUS-RTU protocol. The master controller’s tasks include collecting measured values from the power system nodes, calculating power values for individual converters, selecting the operating state, and setting limits. During tests of the basic algorithm’s correctness, it was crucial to record the instantaneous active power and electrical energy in all power system nodes, as well as the battery SOC. The hydrogen storage status was manually set because of the hydrogen storage simulation. The photovoltaic source generation profile was also simulated.

Active power and energy measurements on the test stand were carried out in three ways:

• Measurements using measurement systems installed on power converters. Each converter is equipped with voltage and current measurement systems, and the converter controller calculates the instantaneous active power;
• Measurements using electricity meters. Electricity meters were installed at the 3 × 400 V AC power grid connection and the output supplying the load. The electricity meters communicate with the master controller via the RS-485 bus and MODBUS-RTU protocol;
• Measurements using the telemetry system built as part of the project. Measurements in DC lines, especially at 3 kV, require a separate measurement system. A telemetry system was built to perform analog voltage, current, and temperature measurements, ensuring galvanic isolation from the master controller. The telemetry system communicates with the master controller via Ethernet.

The power system control layout is shown in Figure 6a, and the telemetry system layout is shown in Figure 6b.

The voltage measurement of the traction grid was carried out with a voltage transducer LEM DVM 4200 (LEM International SA, Meyrin, Switzerland), which has a voltage measurement range of 4200 V. Other DC voltages were measured directly with voltage dividers and a specially designed differential probe. Galvanic isolation has been implemented at the communication level between the measurement module and the main data concentrator. The DC current measurement was carried out using a current transducer LEM LA 100-P (LEM International SA, Meyrin, Switzerland).

The measurement data were recorded in the InfluxDB database and displayed on the Grafana system. The database and Grafana software (Version 8.3.2) were installed on the Raspberry Pi 4 data server. The data server communicated with the master controller via the Ethernet bus, and the measurements were read and saved within a period of 1 s. Figure 7 shows the Grafana data reading window.

2.2. Energy Management Strategy

The control algorithm for the power supply system of the railway signal box is known as the Energy Management Decision Algorithm (EMDA). Its primary goal is to maximize the use of the energy generated by photovoltaic panels and to ensure a continuous power supply to the railway signal box. This is achieved through the efficient management of various energy sources and storage units.

The EMDA operates by forecasting the energy demand of individual components of the system. This forecasting is based on historical operational data and current conditions. The algorithm also includes the ability to predict energy generation from photovoltaic panels, considering weather conditions and the time of day.

Once the forecast is established, the algorithm balances generation and demand. This balance determines whether the energy available from photovoltaic panels is sufficient to meet current demand. In cases where there is a surplus of energy, the algorithm decides to charge the energy storage units, which include batteries and hydrogen cells.

EMDA is designed to monitor the SOC of energy storage units continuously. It makes decisions about which storage units to charge and from which to draw energy to optimize their use in the signal box power supply system. The main priorities are to ensure the uninterrupted operation of the Railway Traffic Control (RTC) devices and to maximize the use of renewable energy. If the energy from photovoltaic panels is insufficient, the algorithm utilizes the energy stored in the storage units. Subsequently, the algorithm considers using energy from the low-voltage grid, and finally from the traction grid, to minimize the load on the main railway infrastructure and reduce the costs associated with energy consumption.

In addition to managing the SOC of the energy storage units, the algorithm controls power flows by setting appropriate power levels on the power converters. These converters handle the conversion and distribution of energy among different energy storage units and the electrical grids. In the event of unavailability or failure of any of the power system units, the algorithm adjusts its management strategy to meet the basic objectives, namely to ensure continuous power supply and maximize the use of renewable energy.

Emergency management is another critical feature of the EMDA. The algorithm is designed to anticipate emergency situations and automatically switch power sources to minimize the risk of power interruptions. In these scenarios, the primary objective is to maintain power supply to the essential RTC devices.

2.3. The Role of Energy Storage Systems in the Railway Signal Box Power Supply System

The main control algorithm’s task is to ensure access to electrical energy using available resources in the event of power grid emergencies. The following resource utilization priorities were adopted in the case of successive power source shutdowns:

-

Photovoltaic source;

-

3 × 400 V AC power grid;

-

3 kV DC traction grid;

-

Hydrogen storage system;

-

Battery storage system.

The primary parameters of the power and energy management system are the SOC of the battery (SOC_AKU) and the hydrogen storage system (SOC_H2). These are natural control signals for the adopted topology without a battery power converter. Power balancing in the system is based on controlling the battery power (limiting charging and discharging currents), and the converter responsible for charging current control will be selected according to the adopted control strategy.

The power balance of the railway signal box power supply system is described by the following formula:

$${\sum }_{i=1}^k{P}_{\mathrm{P}\mathrm{S}\_i}-{\sum }_{i=1}^l{P}_{\mathrm{P}\mathrm{L}\_i}\pm {\sum }_{i=1}^m{P}_{\mathrm{P}\mathrm{E}\mathrm{S}\_i}=0,$$

(1)

where:

• energy sources

$$P_{\mathrm{P}\mathrm{S}}=P_{\mathrm{P}\mathrm{L}1}+P_{\mathrm{P}\mathrm{L}2}+P_{\mathrm{P}\mathrm{G}}+P_{\mathrm{P}\mathrm{P}\mathrm{V}}$$

(2)

loads

$$P_{\mathrm{P}\mathrm{L}}=P_{\mathrm{l}\mathrm{o}\mathrm{a}\mathrm{d}}+P_{\mathrm{a}\mathrm{u}\mathrm{x}}$$

(3)

energy storages

$$P_{\mathrm{P}\mathrm{E}\mathrm{S}}=P_{\mathrm{A}\mathrm{K}\mathrm{U}}+P_{\mathrm{P}\mathrm{H}2}$$

(4)

P_PL1—grid connection power 3 × 400 V AC, P_PL2—traction grid connection power, P_PG—power of the diesel generator, P_PPV—power of the renewable energy source (photovoltaic), P_load—power of receivers supplied from the AC grid, P_AKU—battery power; P_PH2—power of the hydrogen storage; P_aux—auxiliary power of the supply system, the master controller and auxiliary devices (including maintaining the operation of the electrolyzer and fuel cell).

Each energy storage unit has technical limitations in the form of available energy (current SOC) and energy reserve for a specific operating mode. The basic control strategies for the battery storage system are as follows.

• Emergency/backup power—ensuring access to electrical energy for the required duration in case the hydrogen storage system is depleted and power lines L1 and L2 are disconnected;
• Power and energy balancing—temporarily balancing power from the PV source and the hydrogen storage system;
• The basic control strategies for the hydrogen storage system are
• Emergency/backup power—maintaining a predetermined minimum amount of hydrogen for power supply during grid outages;
• Energy balancing—based on the forecasted energy balance in the system, selecting the operating mode of the hydrogen storage system for the next minimum period to either supply or store energy.

It is assumed that in every topology, a battery connected to the 675 V DC bus is present. The power supply system may include a second storage unit in the form of a hydrogen storage system with an electrolyzer and a fuel cell. Due to the characteristics of the hydrogen storage system, it cannot fully participate in power balancing between loads, the PV source, and the power grid. The system is designed to use a single-power electronic converter for both the electrolyzer and the fuel cell, making continuous switching between devices impossible. Each switch between the electrolyzer and the fuel cell requires stabilization of the device’s operation. Therefore, the hydrogen storage system will serve the role of energy balancing and power regulation in predetermined minimum periods (with a specified transition time between operating modes) of either supplying or storing energy.

Battery state-of-charge status check (SOC_AKU). The available battery capacity, indicated by the SOC_AKU coefficient, is divided into 4 fields, as shown in Figure 8a.

Hydrogen storage state-of-charge status check (SOC_H2). The available battery capacity, indicated by the SOC_H2 coefficient, is divided into 4 fields, as shown in Figure 8b.

Explanations for the notation used in Figure 8a:

• SOC_{AKU_max}: Maximum battery energy (fully charged battery), a technical limitation. Charging beyond this value is not possible.
• Area 1 $\left(E_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{m}\mathrm{a}\mathrm{x}}\right)$, $⟨\left.{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{b}\mathrm{m}\mathrm{a}\mathrm{x}},{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{m}\mathrm{a}\mathrm{x}}\right)$: Maximum safe battery energy, an economic limitation. Within this area, further charging is possible but reduces the battery’s lifespan and efficiency. Charging within this area is allowed if the full capacity of the battery is needed, for example, if a power outage from grid lines L1 or L2 is anticipated.
• Area 2 $\left(E_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{b}}\right)$, $\left({\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{r}\mathrm{e}\mathrm{z}},{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{b}\mathrm{m}\mathrm{a}\mathrm{x}}\right)$: Operating area for power balancing mode with renewable energy sources (RES) and fuel cells.
• Area 3 $\left(E_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{r}\mathrm{e}\mathrm{z}}\right)$, $\left({\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{b}\mathrm{m}\mathrm{i}\mathrm{n}},{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{b}\mathrm{m}\mathrm{a}\mathrm{x}}\right)$: Area of available capacity for emergency operation. The available energy should ensure the power supply to the signal box (devices with guaranteed power supply) for at least the minimum accepted time.
• Area 4 $\left(E_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{m}\mathrm{i}\mathrm{n}}\right)$, $\left({\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{m}\mathrm{i}\mathrm{n}},{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{A}\mathrm{K}\mathrm{U}\_\mathrm{b}\mathrm{m}\mathrm{i}\mathrm{n}}⟩\right.$: Minimum safe battery energy, an economic limitation. Within this area, further discharging is possible, but it reduces the battery’s lifespan and efficiency. Discharging within this area is allowed if the full capacity of the battery is needed, for example, if a power outage from grid lines L1 or L2 is anticipated.
• SOC_{AKU_min}: Fully discharged battery, a technical limitation. Discharge beyond this value is not possible.

Explanations for the notation used in Figure 8b:

• ${\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{m}\mathrm{a}\mathrm{x}}$: Maximum energy of the storage (fully charged hydrogen storage), a technical limitation. Charging beyond this value is not possible.
• Area 1 $\left(E_{\mathrm{H}2\_\mathrm{m}\mathrm{a}\mathrm{x}}\right)$, $⟨\left.{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{b}\mathrm{m}\mathrm{a}\mathrm{x}},{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{m}\mathrm{a}\mathrm{x}}\right)$: Maximum safe energy of the storage, an economic limitation. Within this area, further charging is possible, but efficiency decreases. Charging within this area is allowed if the full capacity of the storage is needed, for example, if a power outage from grid lines L1 or L2 is anticipated.
• Area 2 $\left(E_{\mathrm{H}2\_\mathrm{b}}\right)$, $\left({\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{r}\mathrm{e}\mathrm{z}},{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{b}\mathrm{m}\mathrm{a}\mathrm{x}}\right)$: Operating area for power balancing mode with renewable energy sources (RES). The limitation is the switching time between charging and discharging.
• Area 3 $\left(E_{\mathrm{H}2\_\mathrm{r}\mathrm{e}\mathrm{z}}\right)$, $\left({\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{b}\mathrm{m}\mathrm{i}\mathrm{n}},{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{b}\mathrm{m}\mathrm{a}\mathrm{x}}\right)$: Area of available capacity for emergency/backup operation. The available energy should ensure the power supply to the signal box (devices with guaranteed power supply) for at least the minimum accepted time.
• Area 4 $\left(E_{\mathrm{H}2\_\mathrm{m}\mathrm{i}\mathrm{n}}\right)$, $\left({\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{m}\mathrm{i}\mathrm{n}},{\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{b}\mathrm{m}\mathrm{i}\mathrm{n}}\right.⟩$: Minimum safe energy of the storage, an economic limitation. Within this area, further discharging is possible, but efficiency decreases. Discharging within this area is allowed if the full capacity of the storage is needed, for example, if a power outage from grid lines L1 or L2 is anticipated.
• ${\mathrm{S}\mathrm{O}\mathrm{C}}_{\mathrm{H}2\_\mathrm{m}\mathrm{i}\mathrm{n}}$: Fully discharged hydrogen storage, a technical limitation. Discharge beyond this value is not possible.

The hydrogen storage system consists of two devices: a fuel cell and an electrolyzer. Due to technical limitations of the technology, it is not possible to seamlessly switch between charging and discharging, and vice versa. Each of these devices requires a short startup period. Frequent switching between the electrolyzer and the fuel cell is also not economically viable. Therefore, it was decided that the decision on which device to activate would be made hourly, based on the forecasted energy surplus or deficit for the next hour. Figure 9 shows the algorithm used to select the operating mode of the hydrogen storage system according to the SOC.

Explanations of the indicators and dependencies in the algorithm in Figure 9:

$E_{\mathrm{p}\_60\mathrm{m}\mathrm{i}\mathrm{n}}$—forecast 60-min energy balance in the system:

$$E_{\mathrm{p}\_60\mathrm{m}\mathrm{i}\mathrm{n}}=E_{\mathrm{p}\mathrm{P}\mathrm{V}\_60\mathrm{m}\mathrm{i}\mathrm{n}}-E_{\mathrm{p}\mathrm{o}\_60\mathrm{m}\mathrm{i}\mathrm{n}}$$

(5)

where:

• $E_{\mathrm{p}\mathrm{P}\mathrm{V}\_60\mathrm{m}\mathrm{i}\mathrm{n}}$—forecasted hourly energy from the PV source;
• $E_{\mathrm{p}\mathrm{o}\_60\mathrm{m}\mathrm{i}\mathrm{n}}$—forecasted hourly energy demand in the railway signal box.
• If $E_{\mathrm{p}\_60\mathrm{m}\mathrm{i}\mathrm{n}}<0$—forecasted hourly energy deficit. It does not rule out a surplus of temporary power in the analyzed period.
• If $E_{\mathrm{p}\_60\mathrm{m}\mathrm{i}\mathrm{n}}>0$—forecasted hourly energy surplus. It does not rule out a short-term power deficit in the analyzed period.

2.4. Operational Scenarios

2.4.1. Normal Conditions

The system operates with all primary energy sources available. The algorithm prioritizes the use of energy from photovoltaic panels, switching to alternative sources as needed. During normal conditions, the system aims to maximize the utilization of renewable energy while ensuring that the energy storage units remain adequately charged to handle potential fluctuations in energy demand or supply. The algorithm continuously monitors the status of all power sources and adjusts the power distribution to optimize efficiency and minimize costs.

2.4.2. Emergency Conditions

Simulations include scenarios with power outages and failures in primary sources. The algorithm prioritizes the use of energy storage units to ensure continuous power supply. In the event of an emergency, such as a failure of the primary AC grid or the traction grid, the algorithm immediately shifts the load to the energy storage units, including batteries and hydrogen cells. This ensures that critical railway signal box functions remain operational. The algorithm also continuously evaluates the SOC of the storage units and manages their use to extend the available backup power duration until normal conditions are restored or backup power sources, such as diesel generators, are activated.

2.4.3. Variable Conditions

Scenarios with fluctuations in renewable energy generation. The algorithm dynamically balances energy storage and consumption, optimizing system performance. Variable conditions involve changes in the availability of energy from photovoltaic panels due to weather variations or other factors. The algorithm dynamically adjusts the charging and discharging rates of the energy storage units to stabilize the power supply. It ensures that energy is efficiently stored during periods of high generation and appropriately used during low generation periods. By continuously balancing the energy flows, the algorithm maintains optimal performance of the power supply system, ensuring reliable operation even under fluctuating renewable energy conditions.

2.5. Master Controller and Algorithm Implementation

To ensure effective energy management and system reliability, a master controller was implemented to oversee the operation of all converters. This controller receives data from various measurement systems, allowing for precise control and adaptation to changing operational conditions. The algorithm described below was developed with the aim of optimizing energy management in the power system of a railway signal box, considering different topologies and dynamically changing operational conditions. The developed algorithm is very extensive. In addition to standard operating conditions, it also takes into account emergency conditions. The algorithm contains more than 200 sets of final equations of power balances and paths of the power determination sequence in individual power nodes of the system. Due to the volume of the algorithm, it is presented in fragments in the article.

The master controller implements an algorithm for controlling the power system converters of the railway signal box. A different algorithm is implemented for each system topology, and this article will discuss the full topology. Based on information received from telemetry measurement systems, the status of the system is determined. Depending on the status and the forecast for demand and generation, a specific control scenario is executed. The developed algorithm is complex and consists of numerous variables, conditions, and different operating states. For each state, different power balance equations apply. Overall, the algorithm can be divided into three stages.

In the first stage, the value of the power in the system is determined by checking the voltages at the connections of the low-voltage grid and the traction grids are checked, and the state of the power supply system is determined. Four basic power states are distinguished: full, basic, backup, and storage (Figure 10). Depending on determined conditions, the appropriate SOC_HX storage thresholds are set. For storage states, the energy storages include an additional reserve (Area 3), which is also replenished first in the event of a positive energy balance. The system status is indicated by a bit number that defines the system topology at a given moment. For instance, the topology T3A00111011 means the availability of the following system elements: bidirectional PL1 converter, traction grid, photovoltaics, battery, and hydrogen storage tank, while indicating the lack of availability of L1, L2, and the diesel generator.

Each step of the algorithm, including both conditions and system states, is marked with a dedicated symbol. All conditions begin with the letter “C” and are highlighted in red on the diagrams. For example, the symbol C01 represents the first step of algorithm, which involves checking the availability of the primary power source.

In the second stage illustrated in Figure 11, the forecast for demand, generation, and the SOC of the storage systems, including the battery and hydrogen storage, is considered. Based on this, the operating mode of the hydrogen storage is determined, i.e., whether the electrolyzer or the fuel cell is operating. The operation mode of both storage systems, which can act as either a load or a source, is also selected.

In the third phase, the power flow in the system is calculated, and the power is set for individual converters, as illustrated in Figure 12.

2.6. Testing Program and Methodology

In the context of the increasing importance of energy efficiency and the stability of power systems, it is crucial to conduct comprehensive studies that allow for the assessment and optimization of these systems. In this section, we present the methodology and results of studies conducted for a power system designed for a railway control box under various operational scenarios.

A series of tests were conducted to verify the accuracy of the algorithm and the power balancing of the entire system. The tests were performed at lower power levels, where the impact on regulation accuracy is more noticeable due to the greater measurement inaccuracies in the devices. This allowed for a better consideration of power measurement inaccuracies in the inverters. To verify the system’s performance under different conditions, we conducted tests for 60 different scenarios where the following parameters were varied (see

Table 1. Matrix of 60 Scenarios.

Power Supply/Season	Summer	Spring/Autumn	Winter
Only low-voltage grid	HD-S-LV	HD-SA-LV	HD-W-LV
	LD-S-LV	LD-SA-LV	LD-W-LV
	LS-S-LV	LS-SA-LV	LS-W-LV
	HS-S-LV	HS-SA-LV	HS-W-LV
	NG-S-LV	NG-SA-LV	NG-W-LV
Only traction grid	HD-S-TG	HD-SA-TG	HD-W-TG
	LD-S-TG	LD-SA-TG	LD-W-TG
	LS-S-TG	LS-SA-TG	LS-W-TG
	HS-S-TG	HS-SA-TG	HS-W-TG
	NG-S-TG	NG-SA-TG	NG-W-LS
Load sharing between low-voltage and traction grids	HD-S-LS	HD-SA-LS	HD-W-LS
	LD-S-LS	LD-SA-LS	LD-W-LS
	LS-S-LS	LS-SA-LS	LS-W-LS
	HS-S-LS	HS-SA-LS	HS-W-LS
	NG-S-LS	NG-SA-LS	NG-W-LS
Island mode	HD-S-IS	HD-SA-IS	HD-W-IS
	LD-S-IS	LD-SA-IS	LD-W-IS
	LS-S-IS	LS-SA-IS	LS-W-IS
	HS-S-IS	HS-SA-IS	HS-W-IS
	NG-S-IS	NG-SA-IS	NG-W-IS

):

• Demand Profile: The study included three demand profiles corresponding to different seasons. This allowed us to evaluate how the system handles varying seasonal loads.
• PV Generation Profile: Five different PV generation profiles were included, comprising two scenarios of power deficit, two scenarios of power surplus, and one scenario with no generation. This enabled us to analyze the system’s performance under various energy generation conditions.
• Power Distribution among Electrical Connections: We analyzed different power distribution scenarios:
  • Only low-voltage grid connection;
  • Only traction grid connection;
  • No external power supply (island mode);
  • Power split between low-voltage grid and traction grid connections in a 0.5/0.5 ratio.
• Energy Flow Direction at the Connection: We examined two directions of energy flow: energy export to the grid and energy import from the grid.

Notations Used:

• LV—Low-Voltage grid
• TG—Traction grid
• LS—Load Sharing between LV and traction grids
• IS—Island mode (battery, diesel generator, and hydrogen storage)
• PV—Photovoltaic installation
• HD—High Deficit of power generated from PV relative to demand
• LD—Low Deficit of power generated from PV relative to demand
• LS—Low Surplus of power generated from PV relative to demand
• HS—High Surplus of power generated from PV relative to demand
• NG—No Generation
• S—Summer
• SA—Spring/Autumn
• W—Winter

3. Results

As part of the power system tests, we prepared 60 scenarios. The power system was tested in all 60 scenarios. Among the 60 test scenarios, we selected a few that best illustrate the key characteristics of the railway switching station supply system we developed. Figure 13, Figure 14, Figure 15 and Figure 16 present selected short-term profiles of the electric power balance measured in the test supply system shown in Figure 2.

3.1. Selected Short-Term Profiles of Electric Power Balance in the Railway Switching Station Supply System

3.1.1. Scenario LS-S-LV

The supply system includes the following components: a low-voltage power grid, photovoltaic, batteries, a fully charged battery, and a fully charged electrolyzer. Based on the demand forecast, there is a slight surplus of power in the system. Figure 13 shows values of the power of the load, the power of the low-voltage grid connection L1 and the power of the photovoltaic source.

Figure 13. Power balance of LS-S-LV scenario.

Figure 13. Power balance of LS-S-LV scenario.

The profile shows the power balance in the system, the power generated from photovoltaics (PPV), the power demand (Po), and the low-voltage grid power (PL1a). A small surplus of renewable energy generated (approximately 2 kW) is returned to the low-voltage grid, resulting in a negative PL1a value.

The next profile (Figure 14) additionally includes the power value of the battery (PAKU), which should be zero. The observed power difference is due to the lack of battery control, causing it to compensate for the power losses of the PL1 power converter. Other measured powers in the system were zero.

Figure 14. Power balance of LS-S-LV scenario, including the state of charge and power of the energy storage systems.

Figure 14. Power balance of LS-S-LV scenario, including the state of charge and power of the energy storage systems.

Figure 15 shows the unbalanced power graph of LS-S-LV scenario, i.e., the graph of the power value after summing up all the values in the power supply system nodes. Theoretically, the unbalanced power value should be zero, but the PL1 converter draws power for its own needs, which amounts to about 900 W.

Figure 15. Unbalanced power of LS-S-LV scenario.

Figure 15. Unbalanced power of LS-S-LV scenario.

3.1.2. Scenario HD-SA-LS

The supply system includes the following components: a low-voltage power grid, a traction grid, photovoltaic batteries, a fully charged battery, and a half-charged fuel cell. Based on the demand forecast, there is a significant power deficit in the system; i.e., the hydrogen storage works in fuel cell mode. The graphs of individual power values in the system nodes are shown in Figure 16.

Figure 16. Power balance of HD-SA-LS scenario.

Figure 16. Power balance of HD-SA-LS scenario.

The profile shown in the figure illustrates the power balance in the supply system, considering load changes and their impact on the power balance. A step change in the load is visible, resulting in corresponding changes in the power balance:

• For a 5 kW load, power is supplied from the photovoltaic panels (PPV) and the fuel cell (PH2).
• At an 8 kW load, power comes from the PV (PPV), fuel cell (PH2), and battery (PAKU).
• For a 12 kW load, power is supplied from the PV (PPV), H2 (PH2), AKU (PAKU), and the low-voltage grid.
• At a 15 kW load, additional power is used from the traction grid (PL2).

Figure 16 shows noticeable delays in the system’s response to load changes. The average delay is about 53 s, resulting from the inertia of system components and the response times of control algorithms. These delays were systematically minimized during system testing to ensure stable and efficient operation.

Figure 17 shows the unbalanced power graph of the HD-SA-LS scenario, i.e., the graph of the power value after summing up all the values in the power supply system nodes. Theoretically, the unbalanced power value should be zero, but PL1 and PL2 converters charge power for their own needs, which amounts to about 900 W for each other.

3.1.3. Scenario LD-W-IS

The system operates in island mode and consists of the following components: photovoltaic, a half-charged battery, and a half-charged fuel cell. Based on the demand forecast, there is a slight power deficit in the system. At a 5 kW load, energy is mainly supplied by photovoltaic panels (PPV) and the fuel cell (PH2) (Figure 18). The battery practically does not participate in the power supply, maintaining a value close to zero.

In the event of both storages discharging, typical in winter, the system generates a notification about the need for an additional power source, which could be a periodically connected diesel generator. When the load drops to 5 kW, the power supplied by the fuel cell decreases to 3.4 kW and then Po drops to 1.44 kW, the power supplied by the fuel cell decreases to 200 W. Figure 19 shows the unbalanced power for the analyzed scenario. Theoretically, the unbalanced power should be zero in the island mode. However, the system operates the PL1 grid converter, whose auxiliary power is about 900 W. This power was not measured.

The change in load and the system’s response for this configuration occurs within a few seconds, indicating the fuel cell’s lower inertia. Changing the load power causes an imbalance of about two seconds.

Despite the slight power deficit, the system strives to use the available energy resources efficiently. The developed control algorithm ensures a stable power balance and quick response to demand changes. In crisis situations, additional power sources ensure the continuity of the system, which is crucial for its reliability and efficiency.

3.1.4. Scenario HS-SA-LS

The system operates in full supply mode, with access to both the low-voltage grid and the traction grid, as well as both energy storages. The battery is fully charged with an SOC of 80%, while the hydrogen storage is at 50% SOC. Due to the positive generation forecast relative to the demand, the hydrogen storage operates in electrolyzer mode according to the algorithm described in Section 2.2. Figure 20 illustrates several key aspects of energy management in the supply system. Firstly, the electrolyzer is charged at a constant power level due to the surplus power generated by the photovoltaic (PPV) compared to the load power (Po). This prioritization is evident as the electrolyzer receives the excess energy to boost its SOC.

Secondly, the battery is not being charged because its SOC is already at 80%. The system, therefore, prioritizes the electrolyzer due to its lower SOC. Additionally, the diagram shows that the battery helps balance the losses of the PL1 power converter, maintaining system efficiency.

Furthermore, surplus power is initially delivered to the low-voltage grid (L1a) and then to the traction grid (L2). All excess energy from the PPV, which is not used for charging the electrolyzer or the battery, is first directed to the low-voltage grid. As the difference between generated power and demand increases, the remaining surplus is fed into the traction grid.

Finally, when the capacity to deliver surplus power to the grid is exhausted, the power generated by the Ppv is limited. The MPPT (Maximum Power Point Tracking) algorithm is disabled and the power converter P_PV operates in a power limiting mode. This is depicted in the diagram, where the green line (PPV) drops from approximately 3.4 kW to 2.7 kW, indicating that there is no longer any possibility to transfer more energy to either grid or to charge the energy storage. This comprehensive management ensures optimal usage and distribution of the generated energy.

Figure 21 shows the unbalanced power graph. This is the most important parameter in the case of supplying a railway signal box simultaneously from two grids: low-voltage grid and traction grid. The unbalanced power value shows the electric power circular flows between the grid connections. However, it should be remembered that about 900 W are the PL1 converter’s own needs, which are not measured. Therefore, the average value of the loop flow power over a period of 10 s does not exceed 1500 W.

4. Discussion

Multisource power supply systems for buildings are nothing new. Current trends in powering buildings from renewable sources focus on improving energy efficiency and limiting the impact on the power grid by using energy storage, advanced control strategies, and forecasting algorithms. Such solutions strive to provide energy autonomy for buildings, often to some extent limiting the comfort of use and access to energy at all times. Following these trends, the constructed power supply system for the railway signal box must first and foremost provide power to the RTCs. The requirements for these facilities must be met, namely at least two connections to independent power grids, a battery, and a diesel generator. Power supply systems that provide energy to the building use contactors to switch between sources or converters in battery systems. The use of the proposed topology of the constructed power supply system poses other challenges, because the two power grids are connected to each other using bidirectional converters and the power of the converters is set based on the power balance equations implemented in the algorithm and the measured actual values in the system nodes.

Describing the power balance with an equation does not take into account the measurement time and measurement errors. This has a negative impact on the power balance, and unbalanced power appears. In the case of a power supply from one connection or in island mode, the imbalanced power is compensated by the battery. The absence of a battery power converter is an advantage in this case. In the case of a power supply from two power grids, the battery also compensates for the unbalanced power, but only within the system. Applying incorrect power values to the PL1 and PL2 grid converters results in energy flow between these networks. Such states have been identified, and the average power of the circular flows did not exceed 1.5 kW.

Controlling the power of the grid power converters should also take into account the power of the converters’ own needs, which should be measured. In our case, the power measurement in the converter nodes was carried out only on one side of the converter, resulting in compensation for the power of the converter’s own needs from the battery. Therefore, the next step in the work on the algorithm optimization will be to take this fact into account.

Additionally, the analysis of scenarios pointed to the necessity of further refining control algorithms to minimize delays in system response to load changes. In scenarios with significant power deficits, the system exhibited some delays in responding to sudden demand changes, which could impact supply stability. Future research should focus on optimizing response times and integrating more advanced techniques for forecasting energy demand.

The conducted studies confirmed that the designed multisource power supply system for signal boxes can effectively manage available energy resources, ensuring reliable power supply even under challenging operational conditions. This system, through the integration of various energy sources and advanced energy storage systems, demonstrates high flexibility and adaptability to changing conditions.

5. Conclusions

The results of the conducted studies indicate the effectiveness of the designed multisource power supply system for signal boxes in terms of reliability and energy efficiency. This system, which utilizes various energy sources such as the low voltage grid, traction grid, photovoltaic sources, and energy storage systems (batteries and hydrogen), demonstrates the capability to ensure a continuous energy supply under different operational scenarios.

The project used in practice known techniques and topologies of power converter systems to connect two different power grids. The connection between two power grids, a typical AC grid, and a DC traction grid can be found in traction substations, and bidirectional power transmission systems are increasingly used. However, these solutions are used for medium-voltage grids and relatively high power. The designed and built PL1 and PL2 grid converter system allows for truly uninterrupted switching of power between these two networks. Additionally, it is possible to use both networks simultaneously to supply the same railway devices, and the power and direction of energy flow depend on the objective function of the control algorithm. Connecting the battery storage directly to the DC bus of PL1 and PL2 converters reduces losses compared to the typical solution of guaranteed power supply systems and is less prone to failures, as one less converter system was used.

Conducting tests directly in a system close to the real one allowed for immediate identification of the system’s weaknesses. First of all, it was necessary to improve and expand the power measurement systems in the system nodes, as well as to improve communication between the converters and the master controller. In order to reduce the number of measurement systems, the master controller took data from converter measurement systems. The basic task of measurements carried out in the converters is to provide data to the current and voltage regulation systems. As it turned out, the power values calculated on this basis caused an error when calculating the unbalanced power and setting the power value to the converters. This is where the unbalanced power occurs, which is mostly compensated by the battery but also leads to circular flows between the connections to the power grids in the case of power supplies from both grids.

One of the key aspects is the flexibility of the system in responding to changes in energy demand and the generation of energy from renewable sources. Tests carried out in 60 different scenarios, which included variable demand profiles and energy generation, showed that the system can efficiently manage available energy resources. The control algorithm allows for optimal utilization of energy from renewable sources and energy storage, which minimizes operational costs and increases the reliability of energy supply. The built power supply system for the railway signal box allows the use of additional energy sources and energy storages, while, on the other hand, it eliminates the disadvantages of conventional solutions, especially the need to connect a second independent AC grid to the building. To be implemented in production practice, the constructed system must still undergo very rigorous safety tests, which will be the last step before implementation.