Coordinated Control Strategy of Railway Multisource Traction System With Energy Storage and Renewable Energy

Influenced by the growing daily travel demand of citizens and extension of urban land, the construction of the urban railway transit (URT) system is gradually increasing nowadays. This situation implies more electricity consumption in URT and more challenges to URT operation stability. Therefore, based on the existing traction substation, this research proposes a configuration of multi-source traction system (MSTS) for URT with a coordinated control strategy according to power profile of the system. The proposed MSTS, including conventional traction system (CTS), renewable energy source (RES), and energy storage system (ESS), have been modelled first. After that, considering the system dynamic performance and ESS capacity, a coordinated control strategy is designed to manage the MSTS energy flow. The control strategy considers the substation voltage as control signal which is optimized by proposed performance index. The case study is conducted on a three-station railway route with two substations. The proposed MSTS with a coordinated control strategy is compared with normal CTS considering multi-train timetables with different departure intervals. The simulation result illustrates that the proposed MSTS can maintain the state of charge (SOC) of ESS and reduce the substation peak power and voltage fluctuation, which is verified both in short-term and long-term simulation. Meanwhile, it can improve the substation capacity by decreasing the energy consumption of CTS to deal with the increasing demand for URT. the case study denotes that the energy-saving rate can be up to 36.25%, and the peak power is reduced up to 46.32%.

U RBAN railway transit (URT) is the most convenient and safe mode of transportation within cities.However, URT consumes a large quantity of electricity nowadays, as it has become an efficient method to reduce the carbon emission caused by fuel vehicles in most large-scale cities [1], [2].By the end of 2021, 50 cities had been equipped with urban railway systems in China.The total length of the railway was up to 6,988.3 km [3], and this trend of increase in energy consumption will rise continuously with the increasing demand for transportation and the expansion of modern cities [4].Besides, the demand for higher speed and more carrying capacity in the railway system requires much more electricity.Thus, the study on the improvement of energy efficiency should be deeply investigated, especially via renewable energy sources (RES) and energy storage systems (ESS).
One of the technical routes to improving energy efficiency is optimizing the timetable and driving profile of the URT.Timetable optimization increases the overlapping time between accelerating train and braking train and then transfers the regenerative braking energy (RBE) to the traction trains [5].In [6], researchers proposed a cooperative scheduling approach to optimize overlapping time, improving the utilization of RBE.Besides, other objectives are also taken into account with the timetable.In [7], both timetable and passenger waiting time were considered, while the operation cost and timetable were optimized together [8].In terms of driving profile optimization, the main objective is to minimize mechanical energy consumption [9], [10].Researchers optimized train speed profiles based on the maximum-minimum ant system algorithm, reducing the total traction energy in [11].Reference [12] analyzed different approaches such as genetic algorithm, ant colony optimization, and dynamic programming for the optimization of eco-driving method.Furtherly, [13] considers both timetable and speed profiles of train optimization to reduce energy consumption.A similar method can be seen in [14], authors proposed an eco-driving optimization for a single train and timetable optimization for the whole system.Besides, another eco-driving model which considers the uncertainty of climatological conditions is established in [15], reducing 34.7% energy consumption.However, facing the increasing demand for transportation, this timetable and driving optimization method has a limited performance of energy consumption reduction and receptivity improvement of RBE.
ESS provides a potential approach to improve the utilization of RBE and the energy efficiency of URT, via storing and reusing the RBE, which is generally dissipated in brake resistance in URT.Meanwhile, ESS can also decrease the railway system's peak traction power demand.Researchers have proved the validity of RBE reutilization via ESS and its effect on reducing CO 2 emission [16].The siting and sizing optimization of ESS is also a key problem in DC railway system [17].The researchers proposed a cost model to optimize the ESS installation, obtaining a benefit of 188% the initial investment [18].Together with timetable optimization, [19] utilized a nonlinear integer programming model to maximize the utilization of RBE.References [4] and [20] both investigated the URT with ESS and optimized the system configuration via genetic algorithms, which achieved a 15%-30% reduction in traction power consumption.Reference [21] proposed an energy transfer strategy and SOC tracking method, optimizing the discharging threshold of ESS and reducing the peak power of traction substation.Reference [22] investigated the effect of ESS charging threshold on URT energy flow.Reference [23] designed converter control strategy for ESS and optimized dynamic performance of the ESS.Furtherly, [24] proposed distributed cooperative control strategy for URT with ESS based on a value decomposition network, while [25] designed energy management for URT and optimized the charging threshold of ESS by deep reinforcement learning, improving the utilization of RBE.Reference [26] designed a smart control strategy for ESS and RBE in URT system, reducing the energy consumption of substation.However, ESS can only improve the utilization rate of RBE but generate no additional energy to supply the traction substations.
RES like wind power (WP) and photovoltaic (PV) power, which can be installed in railway stations or on the rooftop of train vehicles, is capable of providing traction power or supporting auxiliary facilities [27].Usually, the ESS also plays an indispensable role in URT with RES, eliminating the uncertainty of RES and charged by RES, which requires efficient control strategy and management.A basic structure of DC railway with ESS and RES is proposed in [28], which provides a foundation for further research.In the URT system equipped with ESS and RES, [29] and [30] proposed a day-ahead, minute-ahead, and real-time optimization for the system, which reduces the cost, energy consumption, and peak power of substations.In terms of control strategy, [31] proposed a self-adaptive control method for railway microgrid, including RES and ESS.The ESS will store extra energy from RES and transfer it to railway system when RES power is low.The RES uncertainty will affect the operation stability directly, and ESS charging and discharging actions needs to consider the RES uncertainty, ensuring that ESS has enough SOC to provide traction power.The control strategy of URT system equipped with ESS and RES is still a challenge considering the uncertainty of RES and ESS limited capacity.
Current research has made limited progress on ESS discharging threshold and SOC of ESS in the URT with ESS and RES.Besides, the effort of a multi-train system with different timetables on the URT power flow has not been analyzed.Therefore, based on multi-train URT system with ESS and RES, this study investigates the effect of ESS charging threshold and different departure intervals on the URT.First, a configuration of the multi-source traction system (MSTS), which includes ESS and RES and can be directly installed in the existing traction substation, is proposed in this paper.Then, the mathematical model of URT system, ESS, and RES is introduced.After that, the energy characteristic of proposed MSTS is analyzed, and the control strategy of the MSTS has been designed to ensure the traction power supply for multi-train URT and reduce the peak power of substation.The control strategy is optimized based on the RES historical data or prediction data and traction system operation profile to achieve the better performance of the controller.In the case study, the proposed MSTS with the proposed control strategy is verified with different timetables and compared with normal URT.The main contributions of this research are: 1. Designed a coordinate control strategy for the MSTS, which can reduce the substation energy consumption, peak power, and voltage fluctuation.2. Proposed a performance index to evaluate the substation voltage and power fluctuation and optimized the ESS control to keep the ESS remain a high SOC level.The structure of this article is organized as follows.Section II explains the modeling of the system, including railway system model, ESS model and RES model; in Section III, the power flow analysis method is introduced first, and then the control strategy of the MSTS is designed; case studies based on multi-train system with different departure intervals are conducted in Section IV, and the proposed MSTS is compared with normal traction system via proposed performance index.

II. MODELING OF MULTI-SOURCE TRACTION SYSTEM
The fundamental topology of the proposed MSTS is shown in Fig. 1 [28].The proposed system has four main parts: (1) Railway traction substation.(2) Train vehicle.(3) Energy storage system.(4) Renewable energy.The RES unit consists of wind power and photovoltaic power, which could reduce each other's uncertainties and discontinuity due to different nature (wind power has been applied in URT) [31], [32].The RES and ESS form a micro-grid connected to a DC busbar, and directly linked to the third rail or catenary in the traction substation.This DC busbar allows the ESS and RES to achieve the plug and play function, which has already applied in plenty of projects [33], [34], [35].The control center receives the operation information like power and voltage of substation, power of RES, and SOC of ESS.Then, the control center sends the signal to control charging and discharging of ESS.It is worth noting that the capacity of the original transformer will not be occupied by the additional RES and ESS, which means the capacity of original substation can be improved without modification of transformer.

A. Railway System Model 1) Train Vehicle Motion Model:
The urban transit has strong regularity in daily operation.So, the driving profile, like the velocity, power and position, can be calculated by the timetable and train information.To calculate the power of the train will utilize the method in [36].Single train moving along the rail path is described by ( 1) [15].The resistance force Fr can be calculated by (2).
In the normal operation of urban railway transit, the train vehicle has four primary driving states: accelerating, cruising, coasting, and braking.When the train is accelerating or cruising, it obtains the traction power from the catenary, and generates the regenerative braking energy when it is decelerating.Thus, the traction power demand of the train vehicle and current can be obtained by: 2) Traction Power Network: II-B shows the equivalent circuit model of the railway traction power network.The similar modelling method can be found in [22], [23], and [25].The substation equivalent model includes a voltage source and a resistance, and the voltage of substation U sub can be expressed as follow.
The railway line is modeled as a time-varying resistance, and its value is dependent on the position of train vehicle.The traction train absorbs the power from both substations and the braking train generates the energy for other traction trains or consumes the braking energy in braking resistance.

B. ESS Model
Off-board ESS is usually applied to achieve the receptivity of RBE in the charging mode, and then provide the voltage support for the catenary in the discharging mode.The ESS Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.can be installed in substations or along the railway track.In this study, it is considered to be set in each substation.The off-board ESS has both load and generator functions.
1) ESS Equivalent Circuit Model: The supercapacitor (SC) is considered as the ESS in this research, and its equivalent model is regarded as a resistance R and a capacitance C [24], [25], [37].The voltage of ESS can be calculated by (8), and the state of charge of ESS is shown in (10).
2) ESS Working State: The ESS has three working states: charging, discharging and stand by.The ESS working state is determined by the substation's voltage, and the charging / discharging power of ESS will be limited according to U ESS [38].The ESS power limit for charging and discharging is shown in Fig. 4.
P ESS_chmax and P ESS_dismax represent the maximum value of charging and discharging power of ESS.In the discharging part, the ESS transfers energy to traction trains, and the maximum power of ESS will increase with the voltage of ESS in U 1 < U ESS < U 3 .The ESS will not work (standby) when The mathematic equations of ESS maximum discharging and charging power are expressed in (11).The P ESS_dis is a positive value and the P ESS_ch is a negative value.
To protect the ESS and avoid the deep charging and discharging action, the maximum and minimum values of the SOC of ESS are also considered here, shown in (12).

C. RES Model
The RES unit includes wind power and photovoltaic power, which can eliminate the uncertainty of RES.The control of RES will not be considered in this research, because the objective of this research is not to control and maximize the power of RES.Thus, the RES historical data is utilized here to test the proposed MSTS.
According to the area of a normal traction substation [27], the maximum capacity of RES installed on the substation rooftop can be up to 0.5 MW.Similar to reference [31], the RES capacity is 0.25 MW in this paper, and the RES power will change from 0.05MW to 0.25 MW every 50s to reflect the fluctuation characteristic, which is shown in Fig. 5.The RES 1 and 2 is install in substation 1 and 2, respectively.

A. Power Flow Analysis
To calculate the power flow of the proposed system, the nodal voltage equation method is applied here.Since the proposed system has three stations (including two substations), the system structure can be considered as the one in II-B based on the model described in Section II.The basic method of nodal voltage equation is similar to current research [22], [25],  and the detailed equations regarding accelerating and braking trains are shown in (13) and the related equations for more trains can be obtained as well.
To obtain the single train operation information and multi-train system operation information, the detailed calculation method is divided into two main parts: single-train simulator and multi-train simulator.The single-train simulator, which is shown in (1) to (5), mainly obtains the train traction power demand, train regenerative braking energy, train velocity, and train location via the train movement model.In the multitrain simulator, the above results from single-train simulator will be considered together with timetable information.Then, the power and voltage of substations and catenary can be calculated via nodal voltage method in multi-train simulator.When the number of the train vehicle is more than two or less than two, the power flow can be calculated by increasing or reducing nodes and related nodal voltage equations.The overview of power flow process for the DC traction system is shown in Figure 7: The detailed calculation step is shown below: 1) Input the infrastructure information of train vehicle and traction system to single-train simulator.2) Calculate the train movement information (power, position, velocity) of a single train according to equation ( 1)-(3).3) Determine the departure interval and timetable.4) Initialization of the parameter of traction system.5) Input the single-train operation information, and utilize the nodal voltage equation ( 13) to calculate the system operation information, including power and voltage of substations.6) Output the result.

B. The Operation Modes of MSTS
According to the power demand from the substation, the proposed MSTS has three main operation modes, shown in Fig. 8.In operation mode (a), the ESS is standby, and the traction substation and RES provide traction power for trains, while the RBE is transferred to accelerating trains from braking trains.In operation mode (b), the ESS will discharge to support the train vehicle, because the high-power demand of substation.The traction substation, RBE, ESS and RES provide traction power together.In operation mode (c), the ESS will be charged, and the substation works in a standby mode since the available RBE and RES power supply is higher than the traction power demand.The rest of RES and RBE will be transferred to charge the ESS.It is worth noting that braking trains may be equal or larger than zero and the number of traction trains is equal to or larger than one in operation mode (a) and (b).The number of braking trains may be equal to or larger than one in operation (c).Besides, RES will work all the time which aims to avoid curtailment of the RES power, and the ESS power is limited by the constraint described in section II-B.
A part of RBE will be consumed in braking resistance when there is no train absorbing the energy and the SOC of ESS reaches the maximum value.So, the power of RBE can Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
be expressed as: The traction power demand of train vehicle consists of power from substation and power from ESS and RES and power from RBE.

C. Coordinated Control Strategy
To efficiently manage the power of each unit and reduce the energy consumption CTS, a coordinated control strategy, which considers the substation voltage, RES power, and SOC of ESS, is designed here.The proposed coordinated control strategy aims to reduce the energy consumption and peak power of substation, and stabilize the voltage of substation.Meanwhile, the decrease in RES curtailment is also taken into account.The detailed flow chart of the proposed control strategy is shown in III-D.III-D (a) shows the power flow of the proposed system.The traction substation, ESS and RES will provide the power to URT, while the ESS will be charged by RES and RBE.The work state of each unit is controlled by the proposed coordinated control strategy, and there is no energy exchange between ESS and substation.To be specific, the RES will work all the time.However, the RES power may not be fully utilized when required traction power is lower than the RES power and the SOC of ESS has reached the maximum value.
The proposed coordinated control strategy is shown in Fig. 9 (b).For each time step, the control center will measure the substation voltage, and decide the RES work state.If the substation voltage is larger than open-circuit voltage of substation, the RES will not provide the traction power, while the ESS will be charged by RES and RBE.On the contrary, the RES will participate in the traction power supply when the substation voltage is lower than the open-circuit voltage.Then, decide the ESS work state by comparing the voltage of substation with voltage threshold.The voltage threshold is optimized according to URT operation data, ESS parameter and RES historical data, and the calculation method will be introduced later.The ESS will provide the traction power together with substation and RES when substation voltage is lower than the voltage threshold.If the substation voltage does not reach the threshold value, the ESS will be in a standby state.
In this proposed control strategy, the RES will take part in traction power supply.Besides, ESS can provide traction power to reduce the peak power of substation when substation output power reaches the defined threshold value.As for ESS charging, RES and RBE will transfer energy to ESS when substation does not work in traction mode.Thanks to this strategy, the peak power of substation can be reduced, and the capacity of substation can be improved by adjusting the ESS discharging threshold.Meanwhile, the voltage range of substation will be decreased due to external power source.
It is worthy to note that when ESS discharges and charges, it should satisfy the principle shown in Fig. 4, and the power of ESS will not exceed its maximum value.The ESS will discharge and provide the traction power when the peak power of substation is high.However, the SOC of ESS will be below the minimum value when it provides overmuch power, due to its limited capacity.Thus, the ESS discharge threshold should be considered according to the situation of the whole system.

D. ESS Control Optimization
To achieve a better performance of the proposed coordinated control strategy, the ESS discharging threshold need to be adjusted according to different scenarios.The charging threshold will not be optimized because the SOC of ESS needs to be remained at a high level.The peak power and voltage range of substations are the main indexes of URT operation.Thus, define a performance index J 1 , J 2 , and J 3 shown in (16).α and β are the weight of performance index, and α + β =1: The J 1 , J 2 , and J 3 are shown as follows: P sub_up , U sub_up , and U sub_down are fixed values (shown in Fig. 10) of substation power and voltage, which are chosen as the benchmark of the overshoot and undershoot.The benchmark is chosen according to the traction system information.The power benchmark is chosen by (18), and the voltage benchmarks are chosen as (19) and (20).
where P peak is the peak power of each substation which can be obtained by the conventional traction system simulation, U 4 is the ESS charging trigger voltage.The objective function is set as minimizing the sum of J 1 , J 2 , and J 3 : The particle swarm optimization (PSO) is applied here to optimize the ESS charging threshold, because the PSO has been widely used in parameter optimization problem [39], [44], and the PSO can meet the speed and accuracy requirement.The daily operation information of URT, parameter of ESS, and historical data of RES will be used to optimize the ESS discharging threshold by the above optimization method.The detailed process of optimization will not be discussed here.

IV. CASE STUDY AND ANALYSIS
A. Parameter Setting 1) Topology of the Simulation System: A comparison between the conventional traction system (CTS) and MSTS will be conducted.The topology of simulation system is shown in Fig. 11.This topology which includes three stations with different timetables and multiple trains will test the performance of the proposed MSTS.The CTS has the same structure, but the PV, WP, and ESS are not considered.
2) Main Parameters: The main parameters of urban railway system and ESS are shown in

TABLE III PARAMETERS OF PERFORMANCE INDEX
3) Driving Profile and Timetable: Fig. 12 shows the traction power of the single train, and the velocity of the train is in accelerating, cruising, coasting, and braking mode, respectively.
Both upline and downline have train vehicles operating in the railway system, the simulation period is chosen as 800 seconds.In Fig. 13, two timetables with different departure intervals are applied here, including 120s for peak period and 300s for off-peak period.
Considering the uncertainty of RES, the RES is set as TABLE III, which is similar to the reference [31].The main objective of this research is not to maximize the RES power, so the RES output power is directly connected to the DC bus bar.Both RESs are assumed to change randomly every 50s, and the maximum power is 0.24 MW.
Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.

4) ESS Optimization Parameter Setting:
The benchmark of substation power and voltage is shown in TABLE III, and the weight of performance index is chosen as 0.5 and 0.5, considering power and voltage have same weight.The PSO's population is 50, maximum iteration is 100.

B. Results and Analysis
Based on the setting in Section IV-A, the case study firstly analyzed the substation performance of conventional traction system (CTS) in two different departure intervals.Then, the CTS with ESS and multi-source traction system (MSTS) are compared under the proposed coordinated control strategy.At last, the MSTS is tested in long-term operation, and the system robustness is verified by extreme situation of RES.
1) Scenario A: Power Analysis of CTS: In this scenario, the CTS is investigated under different departure intervals.The comparison between two departure intervals, which stand for the peak period and off-peak period respectively, is conducted here.Fig. 14 illustrates the operation information, including power and voltage of substations in CTS.
In Fig. 14, it can be obtained that both substation 1 and 2 peak power exceed 2.5 MW.The substation power keeps at a high value in peak period, while the substation power has a highest value (3.32MW) in off-peak period.The voltage range of substations is from 812 V to 981 V in peak period, while the range is from 797.85 V to 1000 V in off-peak period.In off-peak period, substations can reach a lower and higher voltage value than the one in peak period, because RBE is consumed in braking resistance rather than being transferred to other trains.
2) Scenario B: Comparison Between CTS With ESS and MSTS: In this scenario, the proposed MSTS with coordinated control strategy is investigated here by comparing the system  dynamic response of the CTS with ESS and.Both peak and off-peak period are considered, and the RES in MSTS is simulated as Fig. 5 which is referred to research [31].The result of ESS discharging threshold optimization is shown in TABLE V.
The operation information of substation 1 and ESS 1 in peak period can be seen in Fig. 15.Due to RES providing a part of traction power, the substation peak power has been furtherly reduced compared to CTS with ESS.The peak power of substation 1 is 1.85MW MSTS, compared with 2.20 MW in CTS with ESS.The voltage drop in MSTS is also less than the one in CTS with ESS, decreasing from 884.84 V to 823.85 V.As for work state of ESS 1, the charging power is larger than the one in CTS with ESS, while it also exports more discharging power, and the SOC of ESS 1 can still be larger than minimal value in MSTS.The substation 2 and ESS 2 has similar result, which is not shown here, and the detailed data for CTS, CTS with ESS and MSTS is summarized in TABLE IV.
Compared with CTS, it can be obtained that the MSTS decreases 16.54% and 36.25%energy consumption from the utility grid during peak and off-peak period, respectively.In terms of proposed performance index J, the proposed MSTS has superior performance among three systems, improving 38.64% in peak period and 94.32% in off-peak.Other detailed comparison between CTS with ESS and MSTS is shown in Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.3) Scenario C: Long-Term Operation Analysis: To further invest the proposed MTST and its coordinated control strategy, a long-term simulation is conducted here from 6:00 to 12:00 (6:00 to 9:00 represents the peak period and 9:00 to 12:00 represents the off-peak period, and departure interval is same as the one shown in Fig. 13).The ESS discharging threshold is optimized respectively according to timetable and RES historical data, and the result is 819.53V (peak period) and 826.29 V (off-peak period).The historical data of RES is considered as The simulation result of substation 1 is shown in Fig. 19, where it shows that the peak power of substation 1 is kept below 2.05 MW and 1.70 MW in peak and off-peak period, respectively.The voltage fluctuation is also remained from 817.9 V to 885.6 V and 826.9 V to 899.98 V in peak and offpeak period.The SOC of ESS 1 is kept above the minimum value which ensure the ESS has enough energy to provide the traction power.As for the power and voltage of substation 2 has the similar result with substation 1, which is not shown here.The energy consumption and performance index comparisons between CTS and MSTS in long-term simulation are conducted here and the result is shown in Fig. 20.The total energy consumption of MSTS is 6839.7 kWh which reduces 22.8 % energy consumption compared with CTS between 6:00 to 12:00.The performance index of MSTS is also decreased 45.1 %.
4) Scenario D: System Robustness Analysis: This section aims to test the sensitivity of proposed MSTS and coordinated control strategy under extreme situations like the output of RES reduces to zero caused by the weather condition or machine fault.Based on the long-term simulation in Section IV-B.3, it is supposed that the output of RES drops to zero for 5 minutes, from 7:00 to 7:05 and from 10:30 to 10:35 in peak and off-peak period, respectively.The operation result of substation 1 and ESS 1 is shown in Fig. 21.
The result shows that the substation 1 peak power and minimum voltage still remain the same when RES drops to zero, because the ESS provides the required power.Regarding the SOC of ESS, a drop from 0.83 to 0.45 can be observed when RES power equals to zero in peak period and a slightly drop (from 0.29 to 0.26) in off-peak period, which ensures that ESS 1 has enough SOC to provide the traction power (Substation 2 and ESS 2 have the similar result which are not Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.shown here).The proposed MSTS has the robustness when RES drops to zero for 5 minutes in peak and off-peak period.
To furtherly investigate the system robustness, it is supposed that the output of RES drops to zero from 7:00 to 7:10 and from 10:30 to 10:40 in peak and off-peak period, respectively.The operation result of substation 1 and ESS 1 is shown in V.
The power of substation 1 increases 21.95 % when RES stops in peak period, while the power increases 2.89 % in off-peak period.The peak of substation is still under the maximum value and reduces to previous states when RES works again.The voltage sag occurs when RES stops, reaching 810.40 V and 825.31V in peak and off-peak period, respectively.As for ESS 1 work state, the SOC of ESS 1 drops to the minimum value when RES stops.However, the SOC can still go back to a high level which reduces the substation peak power.Substation 2 has the similar result which is not shown here.
Thus, the proposed MSTS with the coordinated control strategy can still reduce the peak power and voltage fluctuation and remain the SOC of ESS when facing extreme situations like RES power drops to zero.Namely, the proposed MSTS with the coordinated control strategy has certain robustness.Besides, improving the capacity of ESS can enhance the robustness of system.

V. CONCLUSION
This research proposes a multi-source traction system, which can achieve the plug and play of ESS and RES due to the DC busbar.After building the equivalent model of railway system and ESS, the energy flow calculation method is designed, and the energy distribution is analyzed.Then, a coordinated control strategy is proposed to manage the work state of RES and ESS according to the voltage of substations, including an optimization method of ESS discharging threshold with system performance index.The case study is conducted based on three stations (including two substations), multiple trains, and different departure intervals, considering peak period and off-peak period.The influence of departure interval is analyzed first in the case study, which demonstrates that peak power will increase in the peak period and voltage fluctuation will be more severe in offpeak period.Then, the MSTS is tested in short-term and long-term simulation, compared with CTS equipped with ESS.Both results show the proposed MSTS with the coordinated control strategy can reduce the peak power and voltage sag, and the substation capacity can be improved because of the decreasing power.In long-term simulation, the MSTS save 22.8 % energy consumption from substation compared with CTS.Besides, the robustness of the MSTS is verified by simulating extreme situations like RES fault.The proposed MSTS with coordinated control strategy can be applied in a full railway line, solving with more simulation time and nodal voltage equations.However, the proposed coordinated control strategy needs to be designed separately for scenarios with different departure intervals, and other factors like the effect of the passenger's quantity are not considered.The future research will consider a more detailed modelling method, and the real-time control strategy will be applied to cope with the uncertainty of the system.Besides, ESS will also be charged by substation to achieve an economy operation.At last, the system will be tested in a hardware-in-loop system with actual railway system data.

Fig. 2 .
Fig. 2. The equivalent circuit model of the railway traction power network.

Fig. 6 .
Fig. 6.The equivalent circuit of the railway system.

Fig. 12 .
Fig. 12. Traction and velocity of the train vehicle.

TABLE I PARAMETERS
OF THE RAILWAY SYSTEM

TABLE II PARAMETERS
OF THE ESSS (SUPERCAPACITOR)

TABLE IV RESULT
OF CTS AND CTS WITH ESS AND MSTS Fig. 14.Operation information of substations.

TABLE V OPTIMIZATION
RESULT OF ESS DISCHARGING THRESHOLD

TABLE IV .
The maximum voltages of each substation are same in CTS with ESS and MSTS, because they both have the ESS to absorb the RBE.The result illustrates that the combination of ESS and RES can improve the utilization rate of ESS and reduce the peak power and voltage fluctuation of traction substation.