Advanced and Complex Energy Systems Monitoring and Control: A Review on Available Technologies and Their Application Criteria

Complex energy monitoring and control systems have been widely studied as the related topics include different approaches, advanced sensors, and technologies applied to a strongly varying amount of application fields. This paper is a systematic review of what has been done regarding energy metering system issues about (i) sensors, (ii) the choice of their technology and their characterization depending on the application fields, (iii) advanced measurement approaches and methodologies, and (iv) the setup of energy Key Performance Indicators (KPIs). The paper provides models about KPI estimation, by highlighting design criteria of complex energy networks. The proposed study is carried out to give useful elements to build models and to simulate in detail energy systems for performance prediction purposes. Some examples of energy complex KPIs based on the integration of the Artificial Intelligence (AI) concept and on basic KPIs or variables are provided in order to define innovative formulation criteria depending on the application field. The proposed examples highlight how modeling a complex KPI as a function of basic variables or KPIs is possible, by means of graph models of architectures.


Introduction
Energy systems correct management includes process modeling, process optimization, hardware and software, appropriate setup design, and monitored operation procedures. Scientific and industrial research often addresses the formulation of new energy strategies. When a high number of variables is to be considered, the energy system modeling becomes complex. To this purpose, scaling the model to propose a framework suitable for simulations and measurements related to the effective energy scenario becomes an important issue. A scaled model representing the main scenario of the proposed research is sketched in Figure 1, where two areas can be distinguished: (i) a main area related to the control and management of complex logistics fluxes, big power plants, wide grid networks, and renewable energy sources; (ii) a local area comprising smart cities including smart buildings, local transportation, city lighting, local renewable energy sources, and smart manufacturing energy facilities.
The renewable sources play a very important role in economic and political strategies for energy self-sufficiency of countries. Actually, important technology advances are in the market in terms of biomethane/biogas, photovoltaic, wind, wave, geothermal, hydrogen, thermoelectric, and hydroelectric plants. An important emerging topic for research is the energy harvesting from alternative distributed available sources (light, wind, electromagnetic waves, and vibrations). The full integration of renewable energy sources in complex grid systems takes into account the implementation of sensor and storage systems, and the possibility to apply innovative methodologies for data processing. In this scenario, different systems are matched with conventional and renewable energy sources, storage devices, and efficient loads. The output results of the KPIs drive decisions and procedures such as ordinary and extraordinary maintenance services and in-grid/off-grid remote operations, thus ensuring reliable power and decreasing energy costs.
The paper proposes an overview about energy systems by defining possible variables involved in different energy application fields.

Methodology
The methodology used in this work is sketched in Figure 2 summarizing the following phases: (a) following specifications of research projects some topics concerning energy aspects were extracted; (b) keywords to be used for searching were chosen, such as: Sensors of Measurements, Smart   Artificial Intelligence (AI) algorithms are increasingly used for data processing, thus providing advanced analytical tools to estimate correlations between variables and predicting different scenarios including energy production, load consumption, and risks. Following the model of Figure 1, an analysis can be carried out about possible innovative hardware and software technologies, to be used for energy measurement and for data processing, by identifying possible Key Performance Indicators (KPIs) modeling and simulating complex energy systems. The KPI estimations are strategic to simulate and to optimize the electrical systems, properly using resources, devices, and loads, addressing the network to low-cost solutions and economic risk-mitigation procedures. Models of complex energy systems are usable to simulate the operation of interconnected hybrid micro-grids and in general grid connections in the small, medium, and large period, supporting the choice of possible combinations of equipments and facilities working in a unique system.
In this scenario, different systems are matched with conventional and renewable energy sources, storage devices, and efficient loads. The output results of the KPIs drive decisions and procedures such as ordinary and extraordinary maintenance services and ingrid/off-grid remote operations, thus ensuring reliable power and decreasing energy costs.
The paper proposes an overview about energy systems by defining possible variables involved in different energy application fields.

Methodology
The methodology used in this work is sketched in Figure 2 summarizing the following phases: (a) following specifications of research projects some topics concerning energy aspects were extracted; (b) keywords to be used for searching were chosen, such as: ; (c) searching process over the literature was performed by querying the main international journal databases, especially those focused on energy. The Google Scholar engine was used as well. Open datasets concerning the topics of the examined literature and useful to test AI models were found; (d) the searching process was optimized on a two-step basis: after a pre-screening, some main works were filtered with a particular interest in the most recent ones; this refinement process allowed us to group the selected papers into four classes: (i) sensors, (ii) technology characterization depending on the application fields, (iii) advanced measurement approaches and methodologies, and (iv) energy KPIs; repetitive older papers were neglected; (e) the common basic KPIs related the energy aspects were extracted from the selected papers; (f) criteria were defined to formulate complex KPIs as functions of the basic KPIs or variables.
(c) searching process over the literature was performed by querying the main international journal databases, especially those focused on energy. The Google Scholar engine was used as well. Open datasets concerning the topics of the examined literature and useful to test AI models were found; (d) the searching process was optimized on a two-step basis: after a pre-screening, some main works were filtered with a particular interest in the most recent ones; this refinement process allowed us to group the selected papers into four classes: (i) sensors, (ii) technology characterization depending on the application fields, (iii) advanced measurement approaches and methodologies, and (iv) energy KPIs; repetitive older papers were neglected; (e) the common basic KPIs related the energy aspects were extracted from the selected papers; (f) criteria were defined to formulate complex KPIs as functions of the basic KPIs or variables. The complex KPIs are important to model energy systems characterized by a large number of variables. If (a, b, …) are either significant basic KPIs or measured variables, a complex KPI can be expressed as

Sensor Technologies and Energy Metering Systems
Different technologies can be implemented and executed to measure energy parameters. Smart metering technologies [1][2][3][4][5][6][7][8] are suitable for power quality check, measurements of active and reactive power, optimization of grid control, and power consumption monitoring. Supervisory Control And Data Acquisition (SCADA) systems [9,10] are able to integrate measurement systems by controlling parameters in real time. SCADA systems can be used to set up synoptic dashboards monitoring energy and machine/plant parameters; graphical interfaces are typically used to check temperature, electric power, mismatch losses, voltage peaks, and others. Long Range (LoRa) gateway technology and Zigbee protocols are good candidates for horizontal integration of sensors and networks monitoring energy.
A comfortable technology for measurement of energy efficiency is the infrared thermography [11][12][13][14][15][16] combined with image processing techniques. Sensors can be implemented in complex cloud-connected networks according to the locations of the sites to be controlled. Concerning complex sensor network systems, Zigbee technology [17] and Internet of Things (IoT) devices are appropriate for wireless mesh networks monitoring energy systems. The choice of the network architecture is a function of the data protocol to use and of the data transmission logics. In Table 1, some possible technologies are listed, oriented to sensing and energy metering proposed in the literature and related to the content of this paragraph.

List of Basic KPIs
Research Project Topics The complex KPIs are important to model energy systems characterized by a large number of variables. If (a, b, . . . ) are either significant basic KPIs or measured variables, a complex KPI can be expressed as

Sensor Technologies and Energy Metering Systems
Different technologies can be implemented and executed to measure energy parameters. Smart metering technologies [1][2][3][4][5][6][7][8] are suitable for power quality check, measurements of active and reactive power, optimization of grid control, and power consumption monitoring. Supervisory Control And Data Acquisition (SCADA) systems [9,10] are able to integrate measurement systems by controlling parameters in real time. SCADA systems can be used to set up synoptic dashboards monitoring energy and machine/plant parameters; graphical interfaces are typically used to check temperature, electric power, mismatch losses, voltage peaks, and others. Long Range (LoRa) gateway technology and Zigbee protocols are good candidates for horizontal integration of sensors and networks monitoring energy.
A comfortable technology for measurement of energy efficiency is the infrared thermography [11][12][13][14][15][16] combined with image processing techniques. Sensors can be implemented in complex cloud-connected networks according to the locations of the sites to be controlled. Concerning complex sensor network systems, Zigbee technology [17] and Internet of Things (IoT) devices are appropriate for wireless mesh networks monitoring energy systems. The choice of the network architecture is a function of the data protocol to use and of the data transmission logics. In Table 1, some possible technologies are listed, oriented to sensing and energy metering proposed in the literature and related to the content of this paragraph. Harmonic Distortion (THD) [2][3][4] Voltage in percentage [2]; annual active energy heat view, and nonlinear load analysis [3]; sampling data granularity [4] Simultaneous Application of the clustering and of thermal pixel counting algorithms to the radiometric image enhancing panel defects [12,13] Infrared radiometric temperature [ • C], total energy produced and predicted by ANN [kWh] [12]; infrared radiometric temperature [ • C], % of PV panel variation versus temperature [13] Radiometric image processing of thermal insulation PVC composite panels Evaluation of thermal losses of building panels along the aluminum junctions [14] Infrared radiometric temperature [ • C], homogeneity of aluminum panel junctions (PV) Application in energy router system Applications for monitoring of loads, energy source devices, and energy storage systems [15] Infrared thermometer temperature, load prediction, weather forecasting, calculation of energy needs

Thermal dispersion evaluation in indoor environments
Data mining (k-means algorithm for clustering and the Nearest Neighbor (NN) for classification) enhancing thermal dispersions [16] External temperature, room temperature, classification of parts of thermal image (image processing evaluating the risk of the heat leakage)

Zigbee
Wireless technology able to exchange motion data of human movement in rooms with a centralized air conditioning unit Switching off of centralized air conditioning unit (reducing unused electricity) [17] Display when an area served by an AHU unit is without users, number of empty rooms versus days The main issue for the future research on complex systems will likely be combining smart meter measurements using different sensor technologies with communication networks and protocols, so defining architectures suitable to collect synchronized data for KPI evaluation, and to perform real-time control parameters (such as by SCADA systems monitoring through cloud-connected dashboards).
For what concerns smart buildings, the heating systems are combined with electrical power modules where heating modules could include boilers, cogeneration, heat recovery, and other energy systems to produce heat power. In complex systems, more applications fields are joined, thus increasing the complexity of the model to analyze. In Table 2, the most interesting application fields proposed in the literature are listed. Internet of Things-based systems for greenhouse sensing and actuation [18,19] Temperature, light detection by a photo resistor (measurements in a greenhouse) [18]; monitoring energy consumption and control of photovoltaic generation (to enable powering devices only when needed) [19] Logistics Logistics KPIs based on energy aspects Indicators based on fuel consumption, vehicle kerb, weight, engine stress, maintenance level [20][21][22] Load factor, cargo weight, router length, specific fuel consumption (liters consumed every 100 km), vehicle kerb weight [20,21]; energy and fuel consumption (driver costs) [ Energy production monitoring in industry Energy consumption monitoring in production Multisensor system based on the reading of electrical power consumption of different production machines [33] Power of production machines Other application fields can be found at different scale dimensions with energy being a variable characterizing processes and physical phenomena. The approach to follow to set up KPIs will involve: • a preliminary study to establish the parameters contributing to the energy behavior of the specific application filed; • an interaction analysis of elements in the surrounding environment (for example, buildings, cabling, and lighting contributing to the smart city environment).
The KPIs of complex models can be structured in a multilevel architecture where the KPI of a higher level embeds information of all KPIs of lower levels (the root KPI will represent the final indicator of the whole complex system).

Advanced Measurement Approaches and Methodologies
Measurement approaches and methodologies [34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51], such as sensor allocations and related protocols, mainly involve data processing techniques. Different data analysis tools can be applied to extract more information, optimizing energy systems such as predictions, parameter classifications, and possible unbalanced energy conditions. Supervised and unsupervised AI algorithms represent advanced solutions extracting hidden information and realizing Decision Support Systems (DSSs) for energy management. In Table 3 some methodologies proposed in the literature are listed.  Combining different approaches (for both measuring and processing data) to extract more and new information useful for the definition of new efficient KPIs will be the key concern for researchers in the future.

Energy KPI Indicators
KPIs are fundamental to estimate the energy efficiency of a system and are specific for the application to be considered [52][53][54][55][56][57][58][59][60][61][62][63][64][65][66][67][68][69]. Complex KPIs can be formulated as a combination of more KPIs properly taking into account weights for each parameter. The weights of the variables to assign come from the related importance of the specific KPI. KPIs can refer to energy efficiency, energy quality, economical and business aspects, losses, pollution, consumption, and sustainability. In Table 4, KPIs for energy systems are reported and commented upon.   The proposed state of the art is quite exhaustive about standard indicators including costs, losses, quality, and pollution. Complex systems, such as sustainable energy systems in a large scale (green economy), could require the use of more of these KPIs which can be furthermore interrelated. Tables 1-4 can be associated with each element (subsystem) of the energy system of Figure 1. In Table 5, the references matching the ten subsystems are grouped.  KPIs of complex energy systems are estimated by processing a big quantity of variables. Distributed energy systems require a high computational cost for data processing. In this trend, quantum computing and related frameworks could support this weakness [70]. For the energy applications, another main issue correlated to the data extraction for processing is the communications systems choice which requires optimized networks [71,72]. Quantum computing represents a powerful solution for complex systems data processing when applications deal with fossil, renewable, or nuclear energy, even when different aspects such as energy management, efficiency of innovative materials, grid security, and simulations [73] have to be addressed. Quantum algorithms and quantum computing approaches are also suitable for electrical grid operation planning [74,75]. Energy cloud management [76] and big data analytics [77] become fundamental tools when upgrading to energy data processing issues, especially concerning electricity load forecasting where large datasets are required for modeling self-learning of the AI supervised algorithms.

Basic KPIs and variables discussed in
Pollution monitoring is a research topic too, as carbon dioxide, carbon monoxide, unburned hydrocarbons, particulate matter, sulfur dioxide, and nitrogen oxides emissions have to be counted to match ever higher environmental prescriptions [78][79][80]. To this goal, estimating parameters strictly correlated to the green sustainability indexes becomes of high importance, such as Carbon Footprint (CF) (a parameter taking into account greenhouse gas emissions towards the atmosphere caused, for example, by an energy system in the construction of components, during the operation, and when dismissed). For example, the release of about 6,218,222.4 kg CO2 /year (see Figure 3) can be avoided by installing an 8 MW photovoltaic plant to match the electrical energy needs in the south of Italy. The CF estimation is possible by considering the "factor of emission of the electricity mix" which represents the average value of CO 2 emissions due to the production of electricity in Italy. The factor is provided in Italy by the Ministry of the Environment and is 0.531 kg CO2 /(kWh year).
Concerning renewable energy, data of environmental pollution due to energy generation can be analyzed by means of different tools such as drones (such as for water quality in solar farms by applying underwater image detection [81]), acoustic signal processing in biodiesel production [82], the Life Cycle Impact Assessment (LCIA) approach determining resource consumption and substance release in the environment [83], and a multivariate time series method predicting air pollution [84].
Further important research topics concern the energy storage technologies [85,86], and the related operational approaches [87][88][89][90]. The impact of deep refurbishment and the use of renewable energy sources of buildings can be significant when passing from a single building level to a district scale [91]. In addition, the energy model can be more complex when a capillary distribution in the city is considered for small energy producers such as owners of small wind turbines [92], and hybrid solutions combining solar radiation, wind power, and biomass [93]. Numerical tools and data mining platforms such as Konstanz Information Miner (KNIME) [94][95][96] can support the calculus of complex structured indicators by applying AI data processing. In Appendix A, an example of KNIME data processing predicting PV power is reported.  Concerning renewable energy, data of environmental pollution due to energy generation can be analyzed by means of different tools such as drones (such as for water quality in solar farms by applying underwater image detection [81]), acoustic signal processing in biodiesel production [82], the Life Cycle Impact Assessment (LCIA) approach determining resource consumption and substance release in the environment [83], and a multivariate time series method predicting air pollution [84].
Further important research topics concern the energy storage technologies [85,86], and the related operational approaches [87][88][89][90]. The impact of deep refurbishment and the use of renewable energy sources of buildings can be significant when passing from a single building level to a district scale [91]. In addition, the energy model can be more complex when a capillary distribution in the city is considered for small energy producers such as owners of small wind turbines [92], and hybrid solutions combining solar radiation, wind power, and biomass [93]. Numerical tools and data mining platforms such as Konstanz Information Miner (KNIME) [94][95][96] can support the calculus of complex structured indicators by applying AI data processing. In Appendix A, an example of KNIME data processing predicting PV power is reported.
The monitoring of energy consumption in industrial applications can be optimized by the energy manager who manages data processing and processes correlated to the primary energy consumption [97]. A complex energy system takes into consideration many processes related to energy consumption and production as can happen in industrial applications. In this scenario, process mining implementation (processes automatized by AI controls [98]) could support process implantation and data-driven efficient energy strategies.
Energy Management Systems (EMSs) [99,100] represent important applications and research topics. Different rule-based strategy models are proposed in the literature. Some authors discuss control approach schemes with an operation process for micro-grid systems including forecasting, sensing, and actuation [99]. The energy management problem is typically formulated as a deterministic Optimal Control Problem (OCP) [100].
Other EMS approaches are mainly focused on the analysis of management uncertainties such as fuzzy-based methods, linearization approach, probabilistic method, Monte Carlo method, Gaussian mixture model, estimation distribution and stochastic models [100]. Probabilistic methods are classified as numerical and analytical ones [100]. Hybrid approaches are possible such as scenario based and probabilistic approaches [100]. Control and optimization processes play an important role in EMSs [101].
AI algorithms are proposed as real-time application optimization control algorithms for energy management strategies for hybrid power engines [102], thus suggesting a sim- The monitoring of energy consumption in industrial applications can be optimized by the energy manager who manages data processing and processes correlated to the primary energy consumption [97]. A complex energy system takes into consideration many processes related to energy consumption and production as can happen in industrial applications. In this scenario, process mining implementation (processes automatized by AI controls [98]) could support process implantation and data-driven efficient energy strategies.
Energy Management Systems (EMSs) [99,100] represent important applications and research topics. Different rule-based strategy models are proposed in the literature. Some authors discuss control approach schemes with an operation process for micro-grid systems including forecasting, sensing, and actuation [99]. The energy management problem is typically formulated as a deterministic Optimal Control Problem (OCP) [100].
Other EMS approaches are mainly focused on the analysis of management uncertainties such as fuzzy-based methods, linearization approach, probabilistic method, Monte Carlo method, Gaussian mixture model, estimation distribution and stochastic models [100]. Probabilistic methods are classified as numerical and analytical ones [100]. Hybrid approaches are possible such as scenario based and probabilistic approaches [100]. Control and optimization processes play an important role in EMSs [101].
AI algorithms are proposed as real-time application optimization control algorithms for energy management strategies for hybrid power engines [102], thus suggesting a similar use for a general energy system equipped with an AI supporting decision management. Concerning electricity market bidding, some authors analyze a theoretical framework of energy management optimization, by taking into account the interaction between the Independent System Operator (ISO) agent, commercial user agent, and power plant agent [103].

Conclusions and Perspectives
The paper focused on an overview of technologies, KPI investigation and definition, measurement approaches, and data processing methods, spread out over different energy application fields covering civil and industrial scenarios. The specific literature analysis defines many aspects which have to be considered when more complex systems characterized by multilevel KPIs processing different input parameters are addressed. The present review highlights important elements to be considered in real applications modeling advanced energy systems that manage a large number of variables, including the AI concept improving KPIs or defining new ones.
Complex KPIs can be modeled by architecture based on nodes linked into a unique graph. Each node can represent a Basic Variable (BV), a Basic KPI (BK) over a Complex KPI (CK) formulated as a combination of a BV and BK as in Equation (1). The nodes representing the CK behave as a "supernode" [104]. Each node belongs to a hierarchical level. Different levels represent the whole complex system. The KPI design criteria based on a hierarchical or a multilevel approach allow one to better distinguish the energy efficiency of a single element of the whole energy system. The KPIs characterized by a higher level will contain the information of lower levels. The lower KPIs or variables will be independent from KPIs of higher levels.
A main application field is that of the smart buildings, where energy control and management involve a large number of electric loads and plants, especially if large indoor areas are considered. The formulation of complex KPI systems can define correlated indicators supporting the full energy management process, which can be performed by: • a cloud framework; • reading signals detected by sensors; • processing data by means of AI algorithms predicting daily loads, optimizing energy consumption and loads; • switching electric power (as for energy routing applications).
KPIs in smart buildings could take into account other important aspects such as wellness/security (gas sensing) and can be matched with home automation applications. The KPI model can be more complex if more buildings are considered in the same system to be analyzed; all the KPIs of all the buildings can be combined to define a unique one for a neighborhood or a whole city. The modularity of the model is then useful to scale the application for wider areas. An example of complex KPIs in smart building is provided in Appendix B.
Energy KPI models can also be formulated in particular application fields such as logistics. Actually, logistics applications are commonly characterized by energy aspects. A logistics system can be characterized by different variables contributing directly (vehicle load factor, cargo weight, router length, specific fuel consumption, vehicle kerb weight, etc.) or indirectly (such as for the driver behavior which can influence the vehicle consumption). The KPI model will be useful to optimize logistics fluxes based on the energy behavior model of the fleets. More complex systems can be associated with the joined actions of different vehicles involved in the transportation of the same product (transport by truck, plane, train, ship, etc.), and different logistics networks composed of different hubs. An example of complex KPIs in logistics is provided in Appendix C.
Concerning renewable energy systems, the KPI model can be characterized by different elements such as renewable sources, local electrical networks (medium-voltage electrical cabling and electrical components of the site where the energy sources are allocated), and high-voltage networks. The complexity of the system is increased when different renewable energy fields are considered; the monitoring and control of more PV fields (structured in subfields) transmitting energy to a high-voltage power plant is an example of a complex system. An example of complex KPIs in photovoltaic plants is provided in Appendix D.

Appendix B. Example of Complex KPI in Smart Building
Energy control and management in smart building applications can involve different variables according to loads and plants, especially if large indoor areas are considered. The full energy management process can be performed in a cloud framework, by reading signals detected by sensors, and processing data by means of AI algorithms predicting daily loads, optimizing energy consumption and load and switching electric power (as for energy routing applications). In Figure A2, an example of a complex system integrating data sensors in smart building is shown; the system is able to detect basic energy parameters and to enable loads as a function of data processing. In the proposed example, an efficiency indicator characterizes each floor. In the model, indicators are included estimating photovoltaic and thermal/electrical efficiencies by means of KPIs and different sensor data. All the KPIs of each floor are combined into a high-level KPI representing the energy efficiency of the whole system. The system of Figure A2 takes into account both sensing and actuation functions including AI data processing. All the AI algorithms can be executed into a unique platform as an AI engine. The terms used in the architecture and the meaning of each term are discussed below (BV is a basic value, BK is a basic KPI, CK is a Figure A1. KNIME workflow implementing ANN-MLP algorithm to predict AC power generated by a PV plant. Insets: equivalent ANN-MLP network and predicted average AC power (expressed in kW) processing the dataset [105].

Appendix B. Example of Complex KPI in Smart Building
Energy control and management in smart building applications can involve different variables according to loads and plants, especially if large indoor areas are considered. The full energy management process can be performed in a cloud framework, by reading signals detected by sensors, and processing data by means of AI algorithms predicting daily loads, optimizing energy consumption and load and switching electric power (as for energy routing applications). In Figure A2, an example of a complex system integrating data sensors in smart building is shown; the system is able to detect basic energy parameters and to enable loads as a function of data processing. In the proposed example, an efficiency indicator characterizes each floor. In the model, indicators are included estimating photovoltaic and thermal/electrical efficiencies by means of KPIs and different sensor data. All the KPIs of each floor are combined into a high-level KPI representing the energy efficiency of the whole system. The system of Figure A2 takes into account both sensing and actuation functions including AI data processing. All the AI algorithms can be executed into a unique platform as an AI engine. The terms used in the architecture and the meaning of each term are discussed below (BV is a basic value, BK is a basic KPI, CK is a complex KPI as a function of BV and BK). • Room The complex model of Figure A2 takes into account other important aspects such as wellness/security (gas sensing) and can be matched with home automation applications. The model can be more complex if other buildings are considered in the system to be analyzed. In this case, combining the KPIs of all the buildings, defining a unique one concerning a neighborhood or a city, is also possible. The modularity of the model is then useful to scale up the application to wider areas. The complex KPI referring to the building can be modeled by Equation  The complex model of Figure A2 takes into account other important aspects such as wellness/security (gas sensing) and can be matched with home automation applications. The model can be more complex if other buildings are considered in the system to be analyzed. In this case, combining the KPIs of all the buildings, defining a unique one concerning a neighborhood or a city, is also possible. The modularity of the model is then

Appendix C. Example of Complex KPI in Logistics
Logistics applications are deeply characterized by energy aspects. A logistics system can be characterized by different variables contributing directly or indirectly (such as for the driver behavior which can influence the vehicle consumption) to the estimation of the KPIs. In Figure A3, an example of a complex system associated with estimated KPIs in logistics is illustrated.  (1).

Appendix C. Example of Complex KPI in Logistics
Logistics applications are deeply characterized by energy aspects. A logistics system can be characterized by different variables contributing directly or indirectly (such as for the driver behavior which can influence the vehicle consumption) to the estimation of the KPIs. In Figure A3, an example of a complex system associated with estimated KPIs in logistics is illustrated. Figure A3. Complex architecture in smart logistics transport systems. The graph model with linked nodes was designed by Cytoscope. Specifically, the system of Figure A3 defines a model estimating KPIs in three main hierarchical levels, where the final complex KPI (KPI level 3) is the total KPI ("Energy" KPI) including the calculation of all the parameters contained in the analyzed model. In the example of Figure A3, the KPIs refer to two hypothesized vehicle fleets traveling through two different country regions, and each KPI can be expressed, similar to Equation (1), by the following linear function: where a, b,… are the weight coefficients and y1, y2, … are the parameters/variables defined in Table A1. Each KPI must be properly normalized in order to use consistent scales. Level KPI Description that is the KPI of the driver ( (BV) is the parameter estimating the effect Figure A3. Complex architecture in smart logistics transport systems. The graph model with linked nodes was designed by Cytoscope. Specifically, the system of Figure A3 defines a model estimating KPIs in three main hierarchical levels, where the final complex KPI (KPI level 3) is the total KPI ("Energy" KPI) including the calculation of all the parameters contained in the analyzed model. In the example of Figure A3, the KPIs refer to two hypothesized vehicle fleets traveling through two different country regions, and each KPI can be expressed, similar to Equation (1), by the following linear function: KPI = ay 1 + by 2 + . . .
where a, b, . . . are the weight coefficients and y 1 , y 2 , . . . are the parameters/variables defined in Table A1. Each KPI must be properly normalized in order to use consistent scales. KPI D i that is the KPI of the driver D i (D s (BV) is the parameter estimating the effect of the average velocity provided by a GPS, the revolutions per minute (rpm) (BV) accelerations, and other engine parameters (data provided by the engine control unit); the parameter D e (BV) represents the driver efficiency correlated to a correct driving style (use of AI algorithm).  (1)).

1(CK)
KPI "supernode" embedding information of KPI E level 2 of the two considered fleets: this KPI represents the final "Energy" indicator of the whole complex system.
In Figure A4, an example of the KPI simulation related to the model of Figure A3 is shown. The proposed approach is useful to optimize logistics fluxes, taking into account an efficient energy behavior of the fleets. More complex systems can be associated with the joined actions of different vehicles involved in the transportation of the same product (transport by truck, or by plane, or by train, or by ship, etc.), and different logistics networks composed of different hubs.

Appendix D. Example of Complex KPI in Photovoltaic Plants
Renewable energy systems are characterized by different elements such as renewable sources, local electrical networks (medium-voltage electrical cabling and electrical site components where the energy sources are allocated), and high-voltage networks. The complexity of the system is increased if different renewable energy fields are involved. In Figure A5, an example of a complex system associated with the monitoring and control of two PV fields transmitting energy to a high-voltage power plant is shown. The example refers to the model of two PV controlled fields structured in subfields. Their elements are:   Figure A4. Example of KPI simulation related the model of Figure A3 using Equation (A1) for the calculated KPIs.

Appendix D. Example of Complex KPI in Photovoltaic Plants
Renewable energy systems are characterized by different elements such as renewable sources, local electrical networks (medium-voltage electrical cabling and electrical site components where the energy sources are allocated), and high-voltage networks. The complexity of the system is increased if different renewable energy fields are involved. In Figure A5, an example of a complex system associated with the monitoring and control of two PV fields transmitting energy to a high-voltage power plant is shown. The example refers to the model of two PV controlled fields structured in subfields. Their elements are: The KPIs of the analyzed system are listed in Table A2. All the variables are indicated in the graph of Figure A5.  The KPIs of the analyzed system are listed in Table A2. All the variables are indicated in the graph of Figure A5. Table A2. KPI related to each level of PV complex system to control (BV is a basic value, BK is a basic KPI, CK is a complex KPI as a function of BV and BK).