Traction power supply system control method based on multi-objective optimization

Abstract

The invention provides a traction power supply system control method based on multi-objective optimization, and the method comprises the following steps: S1, determining an initial control variable of a traction system, the initial control variable comprising converter power of an alpha power supply side and a beta power supply side of the traction power supply system and train braking resistor dissipation power of the alpha power supply side, the initial control variable is used as input to construct a regenerative braking energy MPC prediction model; s2, predicting a state variable of a next moment based on a regenerative braking energy MPC prediction model, wherein the state variable comprises output power of a traction substation, total dissipation power of a brake resistor, output power of an energy storage system and a charge state of the energy storage system; and S3, constructing a multi-objective optimization function based on the state variable at the next moment, converting the multi-objective optimization function into a quadratic programming form for solving to obtain a control variable sequence, and taking a first item of the control variable sequence as control input to control the traction power supply system to work.

Description

Technical Field

[0001] The present invention relates to a traction power supply system control method, and in particular to a traction power supply system control method based on multi-objective optimization. Background Art

[0002] The regenerative braking energy management strategy in the railway traction power supply system not only affects the recovery efficiency of RBE (the abbreviation of regenerative braking energy in English), but also determines whether the power borne by the energy storage system is reasonable. It is the key to ensure the safe and stable operation of the train and the energy storage system. The core of the RBE management strategy is to solve the problem of coordinated interaction between the "source-grid-load-storage" of energy when the train is in traction and regenerative braking conditions.

[0003] At present, most regenerative braking energy management strategies only consider the operating cost of the traction power supply system itself or the lowest energy consumption of the train, and pay less attention to the life loss of battery energy storage during the RBE recycling process. When the battery SOC is close to the discharge limit, in order to minimize the power supply cost of the traction substation, the battery is still discharged at a high power, which will accelerate the battery life decay and affect the long-term stable operation of the energy storage system. In addition, the operating conditions of the railway traction system are complex and there are many uncertain factors, making it difficult to accurately control the traction power supply system, which affects the regenerative braking energy recovery and energy storage system. In addition, the power distribution accuracy in the traction power supply system is low, wasting energy.

[0004] Therefore, in order to solve the above technical problems, it is urgent to propose a new technical means. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a traction power supply system control method based on multi-objective optimization, which takes into account the power problems of multiple objectives of source-load-storage in the traction power supply system, adopts model predictive control to predict the operating state of the traction power supply system at future moments, and establishes optimization functions of multiple objectives and converts them into quadratic programming form for solving, and applies the obtained control sequence to the traction power supply system, while taking into account the regenerative braking energy recovery and utilization effect and the SOC state of the battery in the energy storage system, so that the traction power supply system can reasonably allocate the target power of each power unit under various working conditions, reduce the energy consumption of the traction power supply system, and reduce the life loss of the energy storage battery in the traction power supply system.

[0006] The present invention provides a traction power supply system control method based on multi-objective optimization, comprising the following steps:

[0007] S1. Determine the initial control variables of the traction system, which include the converter power of the α power supply side and the β power supply side of the traction power supply system and the power dissipated by the train braking resistor on the α power supply side, and construct a regenerative braking energy MPC prediction model using the initial control variables as input;

[0008] S2. Predicting the state variables at the next moment based on the regenerative braking energy MPC prediction model, the state variables include the output power of the traction substation, the total power dissipated by the braking resistor, the output power of the energy storage system, and the state of charge of the energy storage system;

[0009] S3. Construct a multi-objective optimization function based on the state variables at the next moment, and transform the multi-objective optimization function into a quadratic programming form to obtain a control variable sequence, and use the first item of the control variable sequence as the control input to control the operation of the traction power supply system.

[0010] Furthermore, the regenerative braking energy MPC prediction model is:

[0011]

[0012] Where: P T (k+1) is the output power of the traction substation at time k+1, P R (k+1) is the total power dissipated by the brake resistor at time k+1, P es (k+1) is the output power of the energy storage system at time k+1, SOC es (k+1) is the charge state of the energy storage system at time k+1, Q es.n is the rated capacity of the energy storage system, T s To set the control period, P T (k) is the output power of the traction substation at time k, P R (k) is the total power dissipated by the brake resistor at time k, P es (k) is the output power of the energy storage system at time k, SOC es (k) is the state of charge of the energy storage system at time k, P cα (k) is the converter power on the supply side at time k, P cβ (k) is the converter power on the β supply side at time k, P Rα (k) is the power dissipated by the train braking resistor on the α supply side.

[0013] Further, in step S3, the multi-objective optimization function is:

[0014]

[0015] Where: P q.ave is the average traction power of the train under traction state, P z.ave is the average braking power of the train under braking state, SOC 0 , SOC 1 are the lower and upper limits of the battery's limit charge and discharge range, ρ 1 , 2 ,3 is the weight coefficient and ρ 1 +ρ 2 +ρ 3 =1, N is the total number of prediction time domains.

[0016] Furthermore, the MPC prediction model has the following constraints:

[0017] At time k, the total train load P αβ When (k)≥0, the traction power supply system as a whole presents a traction energy consumption state, and the constraint condition is:

[0018]

[0019] At time k, the total train load P αβ When (k) < 0, the traction power supply system as a whole is in a regenerative braking state, and the constraint condition is:

[0020]

[0021] Where: P es.n The rated power of the energy storage system of the traction power supply system.

[0022] Beneficial effects of the present invention: Through the present invention, the power problem of multiple targets of source-load-storage in the traction power supply system is taken into consideration, model predictive control is adopted to predict the operating state of the traction power supply system at a future moment, and optimization functions of multiple targets are established and converted into quadratic programming form for solving, and the obtained control sequence is applied to the traction power supply system, while taking into account the regenerative braking energy recovery and utilization effect and the SOC state of the battery in the energy storage system, so that the traction power supply system can reasonably allocate the target power of each power unit under various working conditions, reduce the energy consumption of the traction power supply system, and reduce the life loss of the energy storage battery in the traction power supply system. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0024] Figure 1 It is a flow chart of the present invention.

[0025] Figure 2 This is a topological diagram of the regenerative braking energy utilization system of the present invention.

[0026] Figure 3 It is the fuzzy control input / output membership function of the weight coefficient of the present invention.

[0027] Figure 4 The energy storage SOC changes controlled by the fixed weight coefficient and the dynamic weight coefficient.

[0028] Figure 5Comparison of traction substation output power under three optimization objectives.

[0029] Figure 6 Comparison of the total power dissipation of the braking resistor under three optimization objectives.

[0030] Figure 7 The changes in battery energy storage SOC and distribution ranges under three optimization objectives are compared. DETAILED DESCRIPTION

[0031] The present invention is further described in detail below:

[0032] The present invention provides a traction power supply system control method based on multi-objective optimization, comprising the following steps:

[0033] S1. Determine the initial control variables of the traction system, which include the converter power of the α power supply side and the β power supply side of the traction power supply system and the power dissipated by the train braking resistor on the α power supply side, and construct a regenerative braking energy MPC prediction model using the initial control variables as input;

[0034] S2. Predicting the state variables at the next moment based on the regenerative braking energy MPC prediction model, the state variables include the output power of the traction substation, the total power dissipated by the braking resistor, the output power of the energy storage system, and the state of charge of the energy storage system;

[0035] S3. Construct a multi-objective optimization function based on the state variables at the next moment, and convert the multi-objective optimization function into a quadratic programming form for solution, obtain a control variable sequence, and use the first item of the control variable sequence as the control input to control the operation of the traction power supply system. Through the above method, the power problem of multiple objectives of source-load-storage in the traction power supply system is considered, and the model predictive control is used to predict the operating state of the traction power supply system at future moments. An optimization function of multiple objectives is established and converted into a quadratic programming form for solution, and the obtained control sequence is applied to the traction power supply system, while taking into account the regenerative braking energy recovery and utilization effect and the SOC state of the battery in the energy storage system, so that the traction power supply system can reasonably allocate the target power of each power unit under various working conditions, reduce the energy consumption of the traction power supply system, and reduce the life loss of the energy storage battery in the traction power supply system.

[0036] In this embodiment, the topological structure of the traction power supply system is as follows: Figure 2 As shown, the regenerative braking energy MPC prediction model is:

[0037]

[0038] Where: P T (k+1) is the output power of the traction substation at time k+1, P R(k+1) is the total power dissipated by the brake resistor at time k+1, P es (k+1) is the output power of the energy storage system at time k+1, SOC es (k+1) is the charge state of the energy storage system at time k+1, Q es.n is the rated capacity of the energy storage system, T s To set the control period, P T (k) is the output power of the traction substation at time k, P R (k) is the total power dissipated by the brake resistor at time k, P es (k) is the output power of the energy storage system at time k, SOC es (k) is the state of charge of the energy storage system at time k, P cα (k) is the converter power on the supply side at time k, P cβ (k) is the converter power on the β supply side at time k, P Rα (k) is the power dissipated by the train braking resistor on the α supply side.

[0039] According to modern control theory, the general expression of the state equation of a discrete system is

[0040] x(k+1)=Ax(k)+Β 1 u(k)+Β 2 r(k);

[0041] Then: The above prediction model is rewritten as a general expression:

[0042] x(k)=[P T (k) P R (k) P es (k) SOC es (k)] Τ ;

[0043] u(k)=[P cα (k)P cβ (k)P Rα (k)] Τ ;

[0044] r(k)=[P α (k)P β (k)] Τ ;

[0045]

[0046] According to the system state variables at time k+1, the state variables at time k+2 can be predicted as follows:

[0047] x(k+2)=Ax(k+1)+Β 1 u(k+1)+Β 2r(k+1)

[0048] =A 2 x(k)+AB 1 u(k)+AB 2 r(k)+B 1 u(k+1)+B 2 r(k+1)

[0049] Let x(k)=x 0 , and so on, RBE can be expanded to use the MPC model in the system prediction time domain N as

[0050] X k =Mx 0 +C 1 U k +C 2 R k

[0051] in

[0052]

[0053] The MPC prediction model has the following constraints:

[0054] At time k, the total train load P αβ When (k)≥0, the traction power supply system as a whole presents a traction energy consumption state, and the constraint condition is:

[0055]

[0056] In this state, the train needs the energy storage system to discharge and the traction substation to provide energy, and the braking resistor should not be started. Even if there is a small amount of braking power generated by the train on one side of the power supply arm, the braking power should flow to the traction train through the RPC. At the same time, in order to prevent the optimization function from charging the energy storage system by adjusting the energy storage SOC, the maximum output power of the traction substation under this condition is set to the total load power of the train.

[0057] At time k, the total train load P αβ When (k) < 0, the traction power supply system as a whole is in a regenerative braking state, and the constraint condition is:

[0058]

[0059] Where: P es.nis the rated power of the energy storage system of the traction power supply system. In this state, the energy storage system needs to charge and store RBE or the brake resistor starts to consume RBE, and the traction substation does not need to output power at this time. In order to prevent the optimization function from discharging the energy storage system due to the energy storage SOC adjustment, so that the dissipated power of the brake resistor is greater than the braking power of the train, the maximum dissipated power of the brake resistor under this condition is set to the braking power of the train on the same side.

[0060] In this embodiment, under the premise of meeting the train traction load, from the perspective of the source, it is hoped that the traction substation can achieve maximum energy saving; from the perspective of the train load, it is hoped that the train RBE can be fully recovered by the energy storage to reduce the power dissipation of the braking resistor; from the perspective of the energy storage system, it is hoped that the energy storage SOC is far away from the limit charge and discharge range to reduce life loss. Therefore, in step S3, the multi-objective optimization function is:

[0061]

[0062] Where: P q.ave is the average traction power of the train under traction state, P z.ave is the average braking power of the train under braking state, SOC 0 , SOC 1 are the lower and upper limits of the battery's limit charge and discharge range, ρ 1 , 2 , 3 is the weight coefficient and ρ 1 +ρ 2 +ρ 3 =1, N is the total number of prediction time domains. 1 , 2 , 3 They are all dynamic weight coefficients, and their dynamic adjustment process is as follows:

[0063] Let P αβ (k) is the total load power of the train at time k, expressed as P αβ (k) = P α (k)+P β (k).

[0064] According to the total train load P αβ (k) and SOC es (k), five RBEs are designed to utilize the system regulation logic.

[0065] Adjustment logic ①: P αβ (k) is a large braking power and SOC es (k) is relatively high. In order to prevent the optimization function from excessively pursuing to adjust the energy storage SOC to the optimal charge operation range and causing RBE waste, the SOC can be appropriately reduced. es Adjust the weight ρ in the optimization function2 , increase the weight coefficient ρ of the control braking resistor power loss 3 , thereby relaxing the operating requirements for the energy storage SOC, so that the optimization function mainly takes the minimum power loss of the braking resistor as the main goal to fully recover RBE.

[0066] Adjustment logic ②: P αβ (k) is a large braking power and SOC es (k) is relatively low. At this time, the energy storage system should fully charge and recover RBE to keep SOC away from the limit operation range. Therefore, the weight coefficient ρ in the optimization function can be appropriately increased. 2 and ρ 3 , helping to restore the energy storage SOC and reduce the power loss of the braking resistor, and fully recover RBE.

[0067] Adjustment logic ③: P αβ (k) is high traction power and SOC es (k) is relatively high. At this time, the energy storage system has sufficient energy to share the traction load for the traction substation, so the weight coefficient ρ in the optimization function can be appropriately increased. 1 and ρ 2 , so that the energy storage can be fully discharged, the output power of the traction substation can be reduced, and the energy saving rate of the traction substation can be improved.

[0068] Adjustment logic ④: P αβ (k) is high traction power and SOC es (k) is relatively low. At this time, the energy storage system has a small discharge margin. In order to reduce the output power peak of the traction substation, the SOC in the optimization function can be appropriately reduced. es Adjusted weight coefficient ρ 2 , increase the weight coefficient ρ of the traction substation output power control 1 , allowing the energy storage SOC to operate in the limit range for a short time, thereby making full use of the recovered RBE and reducing the output power of the traction substation.

[0069] Adjustment logic ⑤: P αβ (k) is a medium-sized traction braking power and SOC es (k) is in a better charged operation state. At this time, the energy storage system has a better charging and discharging capacity. αβ When (k) is the traction power, it tends to reduce the output power of the traction substation and improve the energy saving rate of the system, that is, increase the weight ρ 1 ; In P αβ When (k) is the braking power, it tends to reduce the braking resistance loss and improve the RBE recovery rate, that is, increase the weight ρ 3 .

[0070] Combination Figure 3 :The current total train load power P αβ(k) and the energy storage system state of charge SOC es (k) is used as the input of the fuzzy controller, and the weight coefficient ρ in the optimization function is 1 , 2 , 3 As the output quantity, the input / output membership function of the fuzzy controller is established. Figure 3 In, P max P αβ The maximum value of (k), P αβ (k) The domain of fuzzy sets is [-P max ,P max ], the corresponding fuzzy subsets are {BZ, MZ, S, MQ, BQ}, which represent large braking power, medium braking power, small power, medium traction power, and large traction power respectively; SOC es (k) The fuzzy domain is [0,1], and the corresponding fuzzy subsets are {VL,L,M,H,VH}, which represent very low, low, medium, high, and very high respectively; the weight coefficient ρ 1 , 2 , 3 The fuzzy domain is [0,1], and the corresponding fuzzy subsets are {VS,S,M,B,VB}, which represent very small, small, medium, large, and very large respectively.

[0071] According to the above fuzzy control rules, ρ in Table 1 and Table 2 is established 1 and ρ 2 Fuzzy control rule table, ρ 3 =1-ρ 1 -ρ 2 .

[0072] Table 1

[0073]

[0074] Table 2

[0075]

[0076] In the above, higher, lower or better are all based on the threshold range of power or state of charge, that is, multiple power threshold ranges and state of charge threshold ranges are set, and different threshold ranges are defined as high, higher, medium, lower, low, etc.

[0077] The multi-objective optimization function is transformed into a quadratic programming form to solve the problem: let ρ′ 1 =ρ 1 / P q.ave , ρ′ 2 =ρ 2 / (0.5-SOC 0 ) 2 , ρ′ 3 =ρ3 / P z.ave , the constant term in the optimization function does not participate in the optimization and can be omitted, so the optimization function can be rewritten as

[0078]

[0079] In the formula, E 1 , A T , A es , A R They are

[0080] E 1(1×N) =[1 1…1]

[0081]

[0082]

[0083]

[0084] X k =Mx 0 +C 1 U k +C 2 R k Substituting into the above formula, omitting the known terms, the optimization function is transformed into a function containing only the control variable U k The standard form of quadratic programming is:

[0085]

[0086] Where H is the quadratic parameter matrix of the control variable to be determined, and f is the linear parameter matrix of the control variable to be determined, which are

[0087]

[0088]

[0089] The constraints of each variable in the prediction time domain can be transformed into the control variable U k is the constraint condition of the parameter, expressed as

[0090] EU k ≤G

[0091] Where E is the coefficient matrix of the linear inequality constraint, G is the right-hand side vector of the linear inequality constraint, which are

[0092]

[0093] Where Y 0 is an N×1 dimensional zero matrix, matrix A Rα , A H , YH , Y es.min , Y es.max , Y T , Y Rα , Y Rβ They are

[0094]

[0095]

[0096]

[0097]

[0098]

[0099]

[0100] In the formula, i=1,2,…,N, j=1.

[0101] The mathematical solution of quadratic programming can be used to obtain the control variable sequence that satisfies the optimization function and constraint conditions in the prediction time domain, that is, the target power P of the converters on both sides of the RPC cα (k), P cα (k+1)…P cα (k+N-1),P cβ (k), P cβ (k+1)…P cβ (k+N-1) and the target power consumption P of the train braking resistor on the α side Rα (k), P Rα (k+1)…P Rα (k+N-1). According to the system power balance relationship, the target reference power that the energy storage system in the RBE utilization system needs to output is calculated.

[0102] In the above, after obtaining the target reference power sequence, only the first item in the reference power sequence is applied to the traction power supply system, and the current state quantity information of the system is re-collected and updated at time k+1, and the optimization is continuously rolled out until the end of the research cycle.

[0103] In this embodiment Figure 4The changes in the energy storage SOC controlled by the fixed weight coefficient and the dynamic weight coefficient are shown. It can be seen that under the control of the MPC energy management strategy with a fixed weight coefficient, the SOC of the energy storage system can be effectively controlled between [0.2, 0.8]; while under the control of the energy management strategy with a dynamic weight coefficient, the energy storage SOC is in the operating range of [0.2, 0.8] most of the time, but in the time periods of 06:43-06:46, 11:00-11:03, 17:02-17:06, the SOC operates in the charging range of [0.8, 0.9], and in the time periods of 10:40-10:45, 16:16-16:22, the SOC operates in the charging range of [0.1, 0.2]. Table 3 shows the RBE recycling effect under the fixed and dynamic weight coefficient energy management strategies within a day. It can be seen that compared with the fixed weight coefficient, the dynamic weight coefficient energy management strategy reduces the power supply of the traction substation by 7.19MWh, increases the energy saving rate by 1.7%, reduces the total energy loss of the brake resistor by 7.18MWh, and increases the RBE recovery rate by 7.67%. This shows that the dynamic weight coefficient adjustment based on fuzzy control in the present invention can relax the control of the energy storage SOC when the train traction energy consumption is high or the RBE is large, and flexibly adjust the output of the energy storage system. Under the premise of ensuring that the energy storage SOC is in the optimal charge range most of the time, the energy saving rate and RBE recovery rate of the traction substation are improved.

[0104] In this embodiment Figure 5 , Figure 6 and Figure 7 The comparison of the output power of the traction substation, the total power dissipated by the brake resistor, and the energy storage SOC under three optimization targets is shown. Among them, MPC optimization target 1 is to optimize the energy storage SOC in the optimal operating range; MPC optimization target 2 is to minimize the output power PT of the traction substation; MPC optimization target 3 is the optimization target of the present invention that takes into account both RBE utilization and energy storage SOC regulation. Figure 5 It can be seen that under the control of the three MPC optimization targets, the PT of the traction substation output power of optimization target 1 is the largest, the PT of optimization target 2 is the smallest, and the PT of optimization target 3 is between the two. That is, in terms of the energy-saving effect of the traction substation, optimization target 2 has the best energy-saving effect, optimization target 3 is the second, and optimization target 1 is the worst. Figure 7 It can be seen that the total dissipated power PR of the brake resistor of optimization target 1 is the largest and occurs in the most time periods, the PR of optimization target 2 is the smallest, and the PR of optimization target 3 is between the two. That is, in terms of the RBE energy storage recovery effect, optimization target 2 is the best, optimization target 3 is the second, and optimization target 1 is the worst. Table 4 shows the RBE recovery effect under the control of the three optimization targets. It can be seen that the RBE recovery effect of optimization target 2 is the best, optimization target 3 is in the middle, and Figure 6 and Figure 7The results are consistent. This shows that the multi-objective optimization function considering source-load-storage in the present invention slightly reduces the energy saving rate and RBE recovery rate of the output traction substation because it will make a trade-off between RBE utilization and energy storage SOC regulation during the adjustment process. Figure 7 By comparing the changes in the battery energy storage SOC and the distribution ranges under the three optimization targets, it can be seen that the energy storage SOC under the control of the three optimization targets is quite different. The energy storage SOC of MPC optimization target 1 operates in the charge range of [0.3, 0.7], and the energy storage charge and discharge depth is low, which is conducive to slowing down the battery life decay. Although MPC optimization target 2 can limit the energy storage SOC between the upper and lower limits of the charge [0.1, 0.9], the energy storage system SOC operates in the low discharge margin range of [0.1, 0.2] for about 41.13% of the time, and the energy storage charge and discharge depth is large, which is easy to accelerate the life loss of the battery, which is not conducive to the long-term stable operation of the energy storage system. Under the control of MPC optimization target 3 in the present invention, the operating time of the energy storage SOC in the interval [0.1, 0.2] is reduced by 31% compared with optimization target 2, and the operating time in the interval [0.2, 0.8] is increased by 30.72%, so the operating range of the energy storage SOC is better. At the same time, compared with optimization target 1, optimization target 3 has a larger SOC operating range, that is, the energy storage system is more fully utilized. In summary, comparing the RBE recycling and utilization effects and energy storage SOC regulation under the control of the three MPC optimization targets, although optimization target 1 has the best energy storage SOC operating range, the RBE loss is too large and the energy saving rate of the traction substation is the lowest; although optimization target 2 recovers the most RBE and the traction substation is the most energy-efficient, the energy storage charged operation state is the worst, which accelerates battery aging; and the optimization target designed by the present invention takes into account both the recycling and utilization of RBE and the energy storage SOC. On the premise of meeting the traction load, it slightly increases the power supply cost of the traction substation, but greatly optimizes the charged operation state of the energy storage and improves the ability of the energy storage to operate continuously and stably.

[0105] Table 3

[0106]

[0107]

[0108] Table 4

[0109]

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A traction power supply system control method based on multi-objective optimization, characterized in that: The following steps are involved: S1. Determine the initial control variables of the traction system, which include the converter power of the α power supply side and the β power supply side of the traction power supply system and the power dissipated by the train braking resistor on the α power supply side, and construct a regenerative braking energy MPC prediction model using the initial control variables as input; S2. Predicting the state variables at the next moment based on the regenerative braking energy MPC prediction model, the state variables include the output power of the traction substation, the total power dissipated by the braking resistor, the output power of the energy storage system, and the state of charge of the energy storage system; S3. Construct a multi-objective optimization function based on the state variables at the next moment, and transform the multi-objective optimization function into a quadratic programming form to obtain a control variable sequence, and use the first item of the control variable sequence as the control input to control the operation of the traction power supply system.

2. The traction power supply system control method based on multi-objective optimization according to claim 1 is characterized in that: The regenerative braking energy MPC prediction model is: Where: P T (k+1) is the output power of the traction substation at time k+1, P R (k+1) is the total power dissipated by the brake resistor at time k+1, P es (k+1) is the output power of the energy storage system at time k+1, SOC es (k+1) is the charge state of the energy storage system at time k+1, Q es.n is the rated capacity of the energy storage system, T s To set the control cycle, P T (k) is the output power of the traction substation at time k, P R (k) is the total power dissipated by the brake resistor at time k, P es (k) is the output power of the energy storage system at time k, SOC es (k) is the state of charge of the energy storage system at time k, P cα (k) is the converter power on the supply side at time k, P cβ (k) is the converter power on the β supply side at time k, P Rα (k) is the power dissipated by the train braking resistor on the α supply side.

3. The traction power supply system control method based on multi-objective optimization according to claim 2 is characterized in that: In step S3, the multi-objective optimization function is: Where: P q.ave is the average traction power of the train under traction state, P z.ave is the average braking power of the train under braking state, SOC0 and SOC1 are the lower and upper limits of the battery limit charge and discharge range respectively, ρ1, ρ2, ρ3 are weight coefficients and ρ1+ρ2+ρ3=1, N is the total number of prediction time domains.

4. The traction power supply system control method based on multi-objective optimization according to claim 2 is characterized in that: The MPC prediction model has the following constraints: At time k, the total train load P αβ When (k)≥0, the traction power supply system as a whole presents a traction energy consumption state, and the constraint condition is: At time k, the total train load P αβ When (k) < 0, the traction power supply system as a whole is in a regenerative braking state, and the constraints are: Where: P es.n Rated power of the energy storage system for the traction power supply system.

Images