An energy storage optimization configuration method for coordinating electric vehicles to participate in the auxiliary service market

Through the least squares support vector machine and CVaR method combined with particle swarm algorithm, the energy storage configuration of electric vehicles participating in the auxiliary service market is optimized, which solves the uncertainty problem of electric vehicles participating in the auxiliary service market, improves the adjustable capacity and reduces scheduling risk.

CN112488362BActive Publication Date: 2024-11-08NORTH CHINA ELECTRIC POWER UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202011228149.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-11-06
Publication Date
2024-11-08
Estimated Expiration
2040-11-06

AI Technical Summary

Technical Problem

How to coordinate the energy storage optimization configuration of electric vehicles participating in the auxiliary service market, improve the adjustable capacity of electric vehicles participating in the auxiliary service market, and solve the uncertainty of electric vehicles participating in the auxiliary service market.

Method used

The least squares support vector machine is used to predict the daily reported capacity of 96 points of electric vehicles, calculate the energy storage demand capacity through Monte Carlo simulation and CVaR, and optimize the energy storage capacity and power configuration in combination with the particle swarm algorithm to minimize the risk of auxiliary service scheduling The objective function of the energy storage optimization configuration.

Benefits of technology

The adjustable capacity of electric vehicles participating in auxiliary services has been improved to ensure system reliability, while minimizing system scheduling risks, and fully leverage energy storage resources and the adjustable potential of electric vehicles.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112488362B_ABST
    Figure CN112488362B_ABST
Patent Text Reader

Abstract

The present invention relates to an energy storage optimization configuration method for coordinating electric vehicles to participate in ancillary service markets, comprising the following steps: step 1, using a least squares support vector machine to predict the daily 96-point reporting capacity of electric vehicles participating in the ancillary service market; fitting to obtain the daily total load distribution of electric vehicles; step 3, determining the error distribution between the reported capacity and the actual response capacity, and using Monte Carlo to simulate the future total daily load capacity scale of electric vehicles; step 4, calculating the daily energy storage demand capacity for coordinating electric vehicles to participate in ancillary services; step 5, constructing an objective function that minimizes the scheduling risk of ancillary services; step 6, solving the energy storage model of step 5 using a particle swarm algorithm, and outputting the configuration results of the optimal energy storage capacity and energy storage power. The present invention can improve the adjustable capacity of electric vehicles participating in ancillary services.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of electric vehicle energy storage optimization configuration, and in particular to an energy storage optimization configuration method for coordinating electric vehicles to participate in an auxiliary service market. Background Art

[0002] As global resource and environmental protection issues become increasingly prominent, the traditional energy flow structure is in urgent need of improvement. my country's power market reform further promotes supply-demand interaction and encourages third-party entities to participate in ancillary services, which is an important means to promote the consumption of new energy and improve energy efficiency in the era of energy Internet. my country's new energy vehicles are developing rapidly. It is estimated that by 2030, the number of electric vehicles (EV) in China will reach 60 million, and the peak charging load will reach 479GW. At the same time, the participation of electric vehicles in the ancillary service market on the demand side has great potential, which plays an important role in promoting the consumption of a high proportion of renewable energy, reducing the cost of ancillary services and promoting load stability.

[0003] The continuous expansion of electric vehicles and the continuous development of V2G (Vehicle to Grid) technology make the research on electric vehicles participating in power auxiliary services of great significance. At the same time, according to the current development stage, there is still great uncertainty in the participation of electric vehicles in the auxiliary service market. Therefore, how to coordinate the optimal configuration of energy storage for electric vehicles participating in the auxiliary service market is a technical problem that technicians in this field need to solve urgently. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for optimizing energy storage configuration for coordinating electric vehicles to participate in ancillary service markets, which can improve the adjustable capacity of electric vehicles participating in ancillary services.

[0005] The present invention solves the technical problem by adopting the following technical solutions:

[0006] A method for optimizing energy storage configuration for coordinating electric vehicles to participate in ancillary service markets comprises the following steps:

[0007] Step 1: Collect the historical load of electric vehicles and use the least squares support vector machine to predict the daily 96-point reported capacity of electric vehicles participating in the auxiliary service market.

[0008] Step 2: The daily total load distribution of electric vehicles is obtained by fitting the historical total load distribution of electric vehicles collected in step 1;

[0009] Step 3: Subtract the daily electric vehicle capacity predicted in step 1 from the actual load value to obtain the response error, and determine the error distribution between the reported capacity and the actual response capacity; based on the daily total load distribution function of electric vehicles obtained in step 2, use Monte Carlo to simulate the response capacity scale of future electric vehicle loads;

[0010] Step 4: Based on the simulation results of the response capacity scale of the future electric vehicle load obtained in step 3, the energy storage demand capacity for coordinating the electric vehicles to participate in the auxiliary services on a daily basis is calculated through CVaR;

[0011] Step 5: Based on the daily 96-point reported capacity curve obtained in step 1 and the daily electric vehicle response capacity scale obtained in step 3, combined with the supplementary difference capacity under the energy storage resource capacity configuration in step 4, construct an objective function to minimize the risk of auxiliary service scheduling, consider the risk loss of different response errors, and achieve optimal configuration of energy storage capacity;

[0012] Step 6: Solve the energy storage model of step 5 using the particle swarm algorithm and output the configuration results of the optimal energy storage capacity and energy storage power.

[0013] Moreover, the collected historical load conditions of electric vehicles in step 1 include annual daily load values ​​and daily 96-point load values.

[0014] Moreover, the specific steps of using the least squares support vector machine in step 1 to predict the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market include:

[0015] (1) Establish a nonlinear mapping The optimal decision function for the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is:

[0016]

[0017] Where w∈R k (k>d) is a high-dimensional feature weight vector, which means the influence of each key influencing factor on the 96-point load value of a single electric vehicle. i ∈R d is the training sample input of 96 load points of d-dimensional electric vehicle, y i ∈R is the training sample output of 96-point load of electric vehicle, b∈R is the bias;

[0018] (2) According to the principle of structural risk minimization, the optimal w and b are found, and the optimization problem of predicting the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is obtained as follows:

[0019]

[0020] Among them, r>0 is the penalty parameter, ξ i is a slack variable, which is a constant that affects the prediction accuracy and can be solved by the PSO algorithm;

[0021] Compared with the standard support vector machine algorithm, there are differences in constraints, as shown in Formula 3;

[0022]

[0023] (3) Applying the Lagrangian function to solve the optimization problem, we get:

[0024]

[0025] Among them, α i is the Lagrange multiplier vector, and α i >0, i=1,2,…,n;

[0026] The final function for predicting the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is shown in formula (5):

[0027]

[0028] in, is the kernel function that satisfies the Mercer condition. Taking the radial basis kernel function as the kernel function, the kernel function formula is shown in formula (6):

[0029]

[0030] Where g is the width coefficient of the kernel function, which is a constant.

[0031] Moreover, the specific method of step 2 is:

[0032] In the process of curve fitting, the annual electric vehicle charging load value data is collected, and samples are selected by random sampling. The samples are fitted with normal distribution, where the normal distribution function is as shown in formula (7):

[0033]

[0034] Then, the partial derivative is taken to obtain the parameter estimation values ​​as shown in formula (8) and formula (9):

[0035]

[0036] Among them, μ and σ are the sample mean and variance of the daily total load of electric vehicles, which are important parameters of the normal distribution function; and are the estimated values ​​of the two parameters μ and σ respectively; N is the number of samples of the total daily load of electric vehicles; x i is the load value of the ith electric vehicle daily total load sample; X is the electric vehicle daily total load sample set.

[0037] Moreover, the specific steps of step 3 include:

[0038] (1) According to the least squares support vector machine prediction method proposed in step 1, the capacity plan reported by electric vehicles participating in the auxiliary service market on the next day is obtained by multiple cycles, and the actual response load value is subtracted to obtain the actual response error, and the error distribution is statistically analyzed;

[0039] (2) fitting the daily total load distribution function of electric vehicles obtained in step 2, and simulating the response capacity scale of electric vehicle loads for 365 days a year through Monte Carlo simulation;

[0040] Among them, the functional form of the fitting function is as follows:

[0041]

[0042] Among them, u is a random number that obeys the uniform distribution of [0,1], is the probability of the daily total load of electric vehicles being distributed in each stage, and z is the response capacity value of the daily total load of electric vehicles that obeys the normal distribution.

[0043] Moreover, the specific method of step 4 is:

[0044] The CVaR difference capacity calculation method based on a certain confidence level can be expressed as formula (13). Under this energy storage capacity agreement, the electric vehicle response capacity is guaranteed to be supplemented at a certain confidence level:

[0045]

[0046] P(f(t,P)≤VaR(p))=1-p (12)

[0047]

[0048] Where, f(t,P) is the shortfall power of the electric vehicle; P t is the reported capacity value of the electric vehicle at time t; p is the confidence level; E(P) is the expected power value of the electric vehicle at time t; P * is the lowest practical response power value that can be provided at time t; is the energy storage capacity of the daily coordinated electric vehicle participation in the auxiliary service market response estimated by CVaR, m is the number of data taken; a is the VaR value; β is the confidence level;

[0049] Moreover, the specific steps of step 5 include:

[0050] (1) In the process of optimizing the energy storage configuration for coordinating electric vehicles to participate in the ancillary service market, in order to minimize the annual dispatch risk of aggregators in the ancillary service market, the objective function of the energy storage optimization configuration model is established (14):

[0051]

[0052] Among them, Pro is the annual scheduling risk calculation formula of the aggregator, which optimizes the energy storage capacity Q through optimization operation:

[0053] Q=Q u +Q l (15)

[0054] (2) Considering that the energy storage power configuration will affect its climbing rate, the constraint function of the energy storage power P is set:

[0055] P ≥ E t ×λ2 (16)

[0056] The calculation method of the parameters is as follows:

[0057] ① Daily risk loss of electric vehicles participating in auxiliary services W d

[0058] After participating in the ancillary service market, electric vehicle companies can obtain ancillary service compensation and compensate for risks through market scheduling, thereby reducing risk losses. The calculation formula for risk losses is shown in (17):

[0059]

[0060]

[0061] Among them, F t K is the peak load auxiliary fee obtained by the market entity during period t; t is the market coefficient, which is the inverse of the average load factor of the thermal power units in the provincial grid. is the peak load contribution rate, E t is the peak load power in period t, is the reference peak load power in period t; t is the response period, is the market clearing price in period t; W is the total daily revenue of auxiliary services; T is the opening period of auxiliary services;

[0062] ② Daily risk cost of electric vehicle assistance

[0063] When electric vehicles participate in auxiliary services, if the difference between the actual valley filling capacity and the reported capacity exceeds the market regulations, there will be opportunity cost losses. The cost calculation method at a single point in time is as shown in formula (19):

[0064]

[0065]

[0066] Among them, A e_t is the penalty cost at time t; γ is the penalty coefficient; is the load deviation rate; λ1 is the penalty critical value;

[0067] The daily risk cost can be obtained by summing up the scheduling costs of 96 points.

[0068] ③ Risk loss of energy storage directly involved in ancillary services

[0069] The calculation formula of direct benefit of ancillary services is shown in formula (20); the benefit of ancillary service electricity price is shown in formula (21); the total risk loss of ancillary service market is shown in formula (22):

[0070]

[0071]

[0072]

[0073] Among them, P battery_u_t is the charging power of the energy storage participating in the ancillary service market at time t; p p and p v They are the energy storage discharge price and charging price respectively; Energy storage capacity to participate in the ancillary services market;

[0074] Among them, E b_battery is the total capacity of the energy storage configuration; the superscript d represents the parameter conditions on the dth day; The ancillary service risk loss of the energy storage system directly participating in the ancillary service at time t; The risk loss of peak and valley electricity prices obtained when energy storage resources are discharged during non-market periods; Total risk of loss for energy storage participating in the ancillary services market; The risk loss of coordinating the energy storage of electric vehicles to participate in ancillary services during period t; To coordinate the peak and valley electricity price risk losses of energy storage participating in ancillary services; To coordinate energy storage to assist electric vehicles and reduce the opportunity risk loss; W l dTo coordinate the total risk loss of energy storage;

[0075] ④ Coordinate the energy storage risk loss W of electric vehicles l d

[0076] The revenue of the ancillary service market of coordinated energy storage is shown in formula (23); the peak-valley electricity price difference revenue of coordinated energy storage is shown in formula (24); the opportunity loss calculation reduced by coordinated energy storage is shown in formula (25); the total revenue of coordinated energy storage is shown in formula (26):

[0077]

[0078]

[0079]

[0080]

[0081] Among them, P battery_l_t To coordinate the charging power of electric vehicles’ energy storage participating in the auxiliary service market at time t; Q l Energy storage capacity to coordinate electric vehicles to participate in auxiliary services; E t_b is the capacity of energy storage auxiliary regulation at time t;

[0082] ⑤Annual cost of energy storage configuration A b

[0083] In the calculation of energy storage configuration cost, the annual value calculation formula of battery energy storage system is used as the energy storage cost calculation method. The calculation method of relevant parameters is as follows:

[0084]

[0085]

[0086] α=λ×β (29)

[0087]

[0088] L eye,D =ηD θ (31)

[0089] Among them, A b is the annual cost of the energy storage system; P bess is the planned energy storage power value; α is the energy storage amortization power cost; E bess is the planned value of energy storage capacity; β is the amortization capacity cost of energy storage; i is the interest rate level; T life is the energy storage service life; C E is the investment cost of the energy storage system; Cm is the maintenance cost of the energy storage system; λ is the ratio of power investment to capacity investment of the energy storage system; K eye,D(i) The depth of charge and discharge is D (i) Battery life of 100 hours; L eye,D is the battery life of the lithium battery when the charge and discharge depth is D; η and θ are important parameters of the charge and discharge life of the lithium battery, which need to be determined based on the experimental fitting results.

[0090] Moreover, the calculation steps of the particle swarm algorithm in step 6 include:

[0091] (1) Parameter determination: Initialize the particle swarm algorithm parameters, including population size, initial position, and initial velocity;

[0092] The initial position X0 and initial speed V0 are randomly generated, and the population size is set to 5; the initial position represents the initial energy storage capacity and power configuration, and the initial speed is the power change step, which is set not to exceed 1;

[0093] (2) Fitness value calculation: Calculate the fitness value of each particle and update the position and velocity of the particle; the fitness value is the objective function in step 5, and the calculation formula is as follows (14); the velocity update formula is formula (32):

[0094] V k+1 = w×V k + c1 × rand() × (p best -X k )+ c2 ×rand()× (g best -X k )(32)

[0095] Among them, V k+1 is the updated speed; w is the inertia factor; V k is the current speed; rand() is a random number generation function that generates a random number between 0 and 1; p best is the optimal position currently achieved by the particle; X k is the current position of the particle; c1 and c2 are learning factors, which learn from the best historical position of the particle and the current best position in the population respectively.

[0096] The position update formula is formula (33).

[0097] X k+1 =X k + V k+1 (33)

[0098] Finally, the optimal energy storage capacity and power configuration can be obtained through optimization results.

[0099] The advantages and positive effects of the present invention are:

[0100] The present invention makes full use of the characteristics of large adjustable capacity and low cost of electric vehicles, strong flexibility and high cost of energy storage resources, and applies the concept of energy storage to coordinate electric vehicles to participate in auxiliary services, which can fully tap the value of energy storage resources while fully tapping the adjustable potential of electric vehicles; the proposed method of scheduling energy storage to participate in auxiliary services based on the principle of minimizing scheduling risk and using the CVaR method can minimize the scheduling risk of the system while ensuring system reliability; the present invention is based on the principle of minimizing scheduling risk, and through coordination strategy research, auxiliary service market capacity determination, scheduling risk determination and optimization calculation, ultimately determines the optimal planning strategy and scheduling results for energy storage to coordinate electric vehicles to participate in the auxiliary service market, thereby improving the adjustable capacity of electric vehicles participating in auxiliary services. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 is a process flow chart of the present invention;

[0102] Figure 2 It is a schematic diagram of the energy storage coordination participation in auxiliary service strategy of the present invention;

[0103] Figure 3 It is a schematic diagram of the relative situation between the reported capacity and the actual response capacity in a specific implementation manner of the present invention;

[0104] Figure 4 is a schematic diagram of simulation results of the capacity scale of electric vehicle load in a specific implementation manner of the present invention;

[0105] Figure 5 (a) is a schematic diagram of the calculation results of daily energy storage reserve capacity in a specific embodiment of the present invention (effective energy storage configuration results);

[0106] Figure 5 (b) is a schematic diagram of the calculation results of daily energy storage reserve capacity in a specific embodiment of the present invention (result of guaranteed energy storage configuration); DETAILED DESCRIPTION

[0107] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings:

[0108] A method for optimizing energy storage configuration to coordinate electric vehicles to participate in the auxiliary service market, such as Figure 1 As shown, the following steps are included:

[0109] Step 1: Collect the historical load of electric vehicles and use the least squares support vector machine to predict the daily 96-point reported capacity of electric vehicles participating in the auxiliary service market.

[0110] In this embodiment, the electric vehicle historical load conditions collected in step 1 include annual daily load values ​​and 96-point daily load values, and the data scale requires at least 365 days of daily load values ​​and 96-point distribution conditions of 30 typical days.

[0111] Support vector machines have been widely used in prediction scenarios and have achieved fruitful results. They have excellent processing capabilities for small sample and high-dimensional data. This is very applicable to situations where there are relatively few quarterly data for multi-energy coupling and the system structure is relatively complex. At the same time, the least squares support vector machine has a significant improvement in prediction accuracy compared to the traditional support vector machine. Therefore, we use the least squares support vector machine (LSSVM) optimized by the particle swarm algorithm (PSO) to predict the daily reported capacity of electric vehicles participating in the auxiliary service market;

[0112] The nonlinear regression support vector machine transforms the data in the input sample space into a high-dimensional linear feature space through a nonlinear transformation through a kernel function, solves the nonlinear problem in the feature space using a linear method, and obtains the global optimal solution. The least squares support vector regression machine replaces the inequality constraints with equality constraints based on the standard support vector machine to speed up the calculation and improve the accuracy.

[0113] In this embodiment, the specific steps of using the least squares support vector machine in step 1 to predict the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market include:

[0114] (1) Establish a nonlinear mapping The optimal decision function for the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is:

[0115]

[0116] Where w∈R k (k>d) is a high-dimensional feature weight vector, which means the influence of each key influencing factor on the 96-point load value of a single electric vehicle. i ∈R d is the training sample input of 96 load points of d-dimensional electric vehicle, y i ∈R is the training sample output of 96-point load of electric vehicle, b∈R is the bias;

[0117] (2) According to the principle of structural risk minimization, the optimal w and b are found, and the optimization problem of predicting the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is obtained as follows:

[0118]

[0119] Among them, r>0 is the penalty parameter, ξ i is a slack variable, which is a constant that affects the prediction accuracy and can be solved by the PSO algorithm;

[0120] Compared with the standard support vector machine algorithm, there are differences in constraints, as shown in Formula 3;

[0121]

[0122] (3) Applying the Lagrangian function to solve the optimization problem, we get:

[0123]

[0124] Among them, α i is the Lagrange multiplier vector, and α i >0, i=1,2,…,n;

[0125] The final function for predicting the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is shown in formula (5):

[0126]

[0127] in, is the kernel function that satisfies the Mercer condition. Taking the radial basis kernel function as the kernel function, the kernel function formula is shown in formula (6):

[0128]

[0129] Where g is the width coefficient of the kernel function, which is a constant;

[0130] Step 2: The daily total load distribution of electric vehicles is obtained by fitting the historical total load distribution of electric vehicles collected in step 1;

[0131] The specific method of step 2 is:

[0132] In the process of curve fitting, the annual electric vehicle charging load value data is collected, and samples are selected by random sampling. The samples are fitted with normal distribution, where the normal distribution function is as shown in formula (7):

[0133]

[0134] Then, the partial derivative is taken to obtain the parameter estimation values ​​as shown in formula (8) and formula (9):

[0135]

[0136]

[0137] Among them, μ and σ are the sample mean and variance of the daily total load of electric vehicles, which are important parameters of the normal distribution function; and are the estimated values ​​of the two parameters μ and σ respectively; N is the number of samples of the total daily load of electric vehicles; x i is the load value of the ith electric vehicle daily total load sample; X is the electric vehicle daily total load sample set.

[0138] Step 3: Subtract the daily electric vehicle capacity predicted in step 1 from the actual load value to obtain the response error, and determine the error distribution between the reported capacity and the actual response capacity; based on the daily total load distribution function of electric vehicles obtained in step 2, use Monte Carlo to simulate the response capacity scale of future electric vehicle loads;

[0139] The specific steps of step 3 include:

[0140] (1) According to the least squares support vector machine prediction method proposed in step 1, the capacity plan reported by electric vehicles participating in the auxiliary service market on the next day is obtained by multiple cycles, and the actual response load value is subtracted to obtain the actual response error, and the error distribution is statistically analyzed;

[0141] (2) fitting the daily total load distribution function of electric vehicles obtained in step 2, and simulating the response capacity scale of electric vehicle loads for 365 days a year through Monte Carlo simulation;

[0142] Among them, the functional form of the fitting function is as follows:

[0143]

[0144] Among them, u is a random number that obeys the uniform distribution of [0,1], is the probability of the daily total load of electric vehicles being distributed in each stage, and z is the response capacity value of the daily total load of electric vehicles that obeys the normal distribution.

[0145] Step 4: Based on the simulation results of the response capacity scale of the future electric vehicle load obtained in step 3, the energy storage demand capacity for coordinating the electric vehicles to participate in the auxiliary services on a daily basis is calculated through CVaR;

[0146] The specific method of step 4 is:

[0147] The CVaR difference capacity calculation method based on a certain confidence level can be expressed as formula (13). Under this energy storage capacity agreement, the electric vehicle response capacity is guaranteed to be supplemented at a certain confidence level:

[0148]

[0149] P(f(t,P)≤VaR(p))=1-p (12)

[0150]

[0151] Where, f(t,P) is the shortfall power of the electric vehicle; P t is the reported capacity value of the electric vehicle at time t; p is the confidence level; E(P) is the expected power value of the electric vehicle at time t; P * is the lowest practical response power value that can be provided at time t; is the energy storage capacity of daily coordinated electric vehicles participating in the auxiliary service market response estimated by CVaR, m is the number of data taken; a is the VaR value; β is the confidence level.

[0152] Step 5: Based on the daily 96-point reported capacity curve obtained in step 1 and the daily electric vehicle response capacity scale obtained in step 3, combined with the supplementary difference capacity under the energy storage resource capacity configuration in step 4, construct an objective function to minimize the risk of auxiliary service scheduling, consider the risk loss of different response errors, and achieve optimal configuration of energy storage capacity;

[0153] The specific steps of step 5 include:

[0154] (1) In the process of optimizing the energy storage configuration for coordinating electric vehicles to participate in the ancillary service market, in order to minimize the annual dispatch risk of aggregators in the ancillary service market, the objective function of the energy storage optimization configuration model is established (14):

[0155]

[0156] Among them, Pro is the annual scheduling risk calculation formula of the aggregator, W d The daily risk loss of electric vehicles participating in auxiliary services on day d; is the daily risk cost of electric vehicle assistance at time t on day d; is the risk loss of energy storage directly participating in ancillary services on day d, W l d Coordinate the energy storage risk loss of electric vehicles for day d; A b The annual cost of energy storage configuration. Through optimization operation, optimize the energy storage capacity Q:

[0157] Q=Q u +Q l (15)

[0158] (2) Considering that the energy storage power configuration will affect its climbing rate, the constraint function of the energy storage power P is set:

[0159] P ≥ E t ×λ2 (16)

[0160] E t is the total charging power of the electric vehicle at time t; λ2 is the critical proportion of charging power of the electric vehicle without participating in valley filling auxiliary service.

[0161] The calculation method of the main parameters is as follows:

[0162] ① Daily risk loss of electric vehicles participating in auxiliary services W d

[0163] After participating in the auxiliary service market, electric vehicle companies can obtain auxiliary service compensation, and compensate for risks through market scheduling, thereby reducing risk losses. The present invention takes the North China auxiliary service market as an example, and the calculation formula for its risk loss is shown in (17):

[0164]

[0165]

[0166] Among them, F t K is the peak load auxiliary fee obtained by the market entity during period t; t is the market coefficient, which is the inverse of the average load rate of the thermal power units in the provincial grid. The present invention temporarily sets it to 1.2; is the peak load contribution rate, E t is the peak load power in period t, is the reference peak-shaving power in time period t; t is the response time period, which is 15 minutes in the present invention; is the market clearing price in period t; W is the total daily revenue of auxiliary services; T is the opening period of auxiliary services;

[0167] ② Daily risk cost of electric vehicle assistance

[0168] When electric vehicles participate in auxiliary services, if the difference between the actual valley filling capacity and the reported capacity exceeds the market regulations, there will be opportunity cost losses. The cost calculation method at a single point in time is as shown in formula (19):

[0169]

[0170] Among them, A e_t is the penalty cost at time t; γ is the penalty coefficient; is the load deviation rate; λ1 is the penalty critical value;

[0171] The daily risk cost can be obtained by summing up the scheduling costs of 96 points.

[0172] ③ Risk loss of energy storage directly involved in ancillary services

[0173] like Figure 2 The schematic diagram of energy storage coordination and participation in auxiliary service strategy is shown in Figure 1, where E b_battery is the total capacity of the energy storage configuration; the superscript d represents the parameter conditions on the dth day; The ancillary service risk loss of the energy storage system directly participating in the ancillary service at time t; The risk loss of peak and valley electricity prices obtained when energy storage resources are discharged during non-market periods; Total risk of loss for energy storage participating in the ancillary services market; The risk loss of coordinating the energy storage of electric vehicles to participate in ancillary services during period t; To coordinate the peak and valley electricity price risk losses of energy storage participating in ancillary services; To coordinate energy storage to assist electric vehicles and reduce the opportunity risk loss; W l d To coordinate the total risk loss of energy storage;

[0174] Among them, the calculation formula of direct income of ancillary services is shown in formula (20); the income of ancillary service electricity price is shown in formula (21); and the total risk loss of the ancillary service market is shown in formula (22):

[0175]

[0176]

[0177]

[0178] Among them, P battery_u_t is the charging power of the energy storage participating in the ancillary service market at time t; p p and p v They are the energy storage discharge price and charging price respectively; Energy storage capacity to participate in the ancillary services market;

[0179] ④ Coordinate the energy storage risk loss W of electric vehicles l d

[0180] The revenue of the ancillary service market of coordinated energy storage is shown in formula (23); the peak-valley electricity price difference revenue of coordinated energy storage is shown in formula (24); the opportunity loss calculation reduced by coordinated energy storage is shown in formula (25); and the total revenue of coordinated energy storage is shown in formula (26).

[0181]

[0182]

[0183]

[0184]

[0185] Among them, P battery_l_t To coordinate the charging power of electric vehicles’ energy storage participating in the auxiliary service market at time t; Q l Energy storage capacity to coordinate electric vehicles to participate in auxiliary services; E t_b is the capacity of energy storage auxiliary regulation at time t;

[0186] ⑤Annual cost of energy storage configuration A b

[0187] In the calculation of energy storage configuration cost, the annual value calculation formula of battery energy storage system is used as the energy storage cost calculation method. The calculation method of relevant parameters is as follows:

[0188]

[0189]

[0190] α=λ×β (29)

[0191]

[0192] L eye,D =ηD θ (31)

[0193] Among them, A b is the annual cost of the energy storage system; P bess is the planned energy storage power value; α is the energy storage amortization power cost; E bess is the planned value of energy storage capacity; β is the amortization capacity cost of energy storage; i is the interest rate level; T life is the energy storage service life; C E is the investment cost of the energy storage system; C m is the maintenance cost of the energy storage system; λ is the ratio of power investment to capacity investment of the energy storage system; K eye,D(i) The depth of charge and discharge is D (i) Battery life of 100 hours; L eye,D is the battery life of the lithium battery when the charge and discharge depth is D; η and θ are important parameters of the charge and discharge life of the lithium battery, which need to be determined based on the experimental fitting results.

[0194] Step 6: Solve the energy storage model of step 5 using the particle swarm algorithm and output the configuration results of the optimal energy storage capacity and energy storage power.

[0195] The calculation steps of the particle swarm algorithm in step 6 include:

[0196] (1) Parameter determination: Initialize the particle swarm algorithm parameters, including population size, initial position, initial velocity, etc.

[0197] Randomly generate the initial position X0, initial speed V0, etc., and set the population size to 5. The initial position represents the initial energy storage capacity and power configuration, and the initial speed is the power change step, which is set not to exceed 1;

[0198] (2) Fitness value calculation: Calculate the fitness value of each particle and update the position and velocity of the particle; the fitness value is the objective function in step 5, and the calculation formula is as follows (14); the velocity update formula is formula (32):

[0199] V k+1 = w×V k + c1 × rand() × (p best -X k )+ c2 ×rand()× (g best -X k )(32)

[0200] Among them, V k+1 is the updated speed; w is the inertia factor; V k is the current speed; rand() is a random number generation function that generates a random number between 0 and 1; p best is the optimal position currently achieved by the particle; X k is the current position of the particle; c1 and c2 are learning factors, which learn from the best historical position of the particle and the current best position in the population respectively.

[0201] The position update formula is formula (33).

[0202] X k+1 =X k + V k+1 (33)

[0203] Finally, the optimal energy storage capacity and power configuration can be obtained through optimization results.

[0204] The present invention is further described below by means of specific examples:

[0205] Step 1: Collect the historical load information of electric vehicles and determine the daily reporting capacity calculation plan for electric vehicles participating in the auxiliary service market.

[0206] Collect the historical load conditions of electric vehicles, including the annual daily load values ​​and the daily 96-point load values. The data scale requires at least 365 days of daily load values ​​and 96-point distribution conditions of 30 typical days. Determine the calculation method of the daily reported capacity of electric vehicles participating in the auxiliary service market, that is, the specific scheme for prediction through time series or big data methods. The present invention adopts the support vector machine method for prediction.

[0207] Collect the response capacity data of electric vehicle load in Beijing in 2019. The model predicts the response capacity through the support vector machine model and uses it as the daily capacity reporting plan. The relative situation between the actual response capacity and the reported capacity curve on a certain day is shown in the figure below. Figure 3 shown.

[0208] Step 2: Fit the electric vehicle load distribution using the historical electric vehicle load distribution collected in step 1.

[0209] In the process of curve fitting, the annual electric vehicle charging load value data is collected, and samples are selected by random sampling. The samples are fitted with normal distribution, where the normal distribution function is as shown in formula (1).

[0210]

[0211] Then, the partial derivative is taken to obtain the parameter estimation values ​​as shown in formula (2) and formula (3).

[0212]

[0213]

[0214] Among them, μ and σ are important parameters of the normal distribution function. and are their estimated values ​​respectively; N is the number of load samples; x i is the load value of the i-th sample; X is the electric vehicle load sample set.

[0215] That is, the annual distribution of electric vehicle load can be obtained.

[0216] During the fitting process, the response scale of the ancillary service market for 100 days was randomly sampled to fit the distribution function. The results showed that the daily response capacity scale obeyed a normal distribution with a mean of 368.88 and a variance of 24.34.

[0217] Step 3: Calculate the capacity of daily electric vehicles participating in the auxiliary service market using the capacity calculation method of step 1, and subtract the calculated result from the actual load value to obtain the response error, and determine the error distribution between the reported capacity and the actual response capacity; use the electric vehicle load distribution function obtained in step 2 to simulate the capacity scale of future electric vehicle loads using Monte Carlo.

[0218] According to the support vector machine prediction method proposed in step 1, the capacity plan reported by electric vehicles participating in the auxiliary service market on the next day is obtained by multiple cycles, and the actual response error is obtained by subtracting it from the actual response load value, and the error distribution is statistically analyzed. According to the fitting function obtained in step 2, the electric vehicle response capacity scale for 365 days a year is simulated through Monte Carlo.

[0219] The inverse function is shown in formula (4).

[0220]

[0221] Among them, u is a random number that obeys the uniform distribution [0,1], and z is the response capacity value that obeys the normal distribution.

[0222] Based on the response capacity reporting scheme obtained in step 1, 960 daily load curves are calculated, and the distribution of response deviations is shown in Table 1.

[0223] Table 1 Distribution of deviations between reported capacity and actual response capacity

[0224]

[0225] Based on the normal distribution function obtained in step 2 and the normrnd function in MATLAB, the capacity scale of electric vehicles participating in the market for 365 days is simulated. The results are as follows Figure 4 shown. Figure 4 The cross punctuation is the daily market size of electric vehicles participating in auxiliary services.

[0226] Step 4: Based on the simulation results of the response capacity scale of the electric vehicle load, the daily energy storage demand capacity for coordinating the auxiliary services of electric vehicles is calculated through CVaR.

[0227] The CVaR difference capacity calculation method based on a certain confidence level can be expressed as formula (8). Under this energy storage capacity agreement, the electric vehicle response capacity can be guaranteed to be supplemented at a certain confidence level;

[0228]

[0229] P(f(t,P)≤VaR(p))=1-p (7)

[0230]

[0231] Where, f(t,P) is the shortfall power; P t is the actual response power value required at time t; p is the confidence level; E(P) is the expected power value at time t; P * is the lowest practical response power value that can be provided at time t; is the energy storage capacity estimated by CVaR for daily coordination of electric vehicles participating in the auxiliary service market response, m is the number of data taken; a is the VaR value; β is the confidence level.

[0232] According to the load simulation results, we first take a confidence level of 90% and use the CVaR method to calculate the daily peak load reserve energy storage capacity. Define security energy storage and efficiency energy storage, where security energy storage is the energy storage capacity configured to keep the electric vehicle load error within the market critical line; efficiency energy storage is the energy storage capacity configured to make the electric vehicle load error as close to zero as possible. The distribution of the calculation results is as follows: Figure 5 As shown, Figure 5 (a) is the result of efficient energy storage configuration, that is, in this case, it can ensure that the load curve after adjustment can completely correspond to the load curve. Figure 5 (b) is the result of guaranteed energy storage configuration, that is, in this case, the load error after adjustment can be within the allowable range. In this example, when the energy storage is sufficient, the efficient energy storage configuration method is preferred; when the energy storage is insufficient, the guaranteed energy storage configuration method is selected; when the requirements cannot be met, the energy storage chooses to directly participate in the auxiliary service market and is not responsible for adjusting the electric vehicle load curve.

[0233] Step 5: Based on the daily reported capacity curve obtained in step 1 and the daily electric vehicle response capacity scale obtained in step 3, combined with the supplementary difference capacity under the energy storage resource capacity configuration in step 4, construct an objective function that minimizes the risk of auxiliary service scheduling, consider the risk loss of different response errors, and achieve optimal configuration of energy storage capacity.

[0234] In the process of optimizing the energy storage configuration for coordinating electric vehicles to participate in the ancillary service market, an objective function (28) is established to minimize the annual dispatch risk of aggregators in the ancillary service market.

[0235]

[0236] Among them, Pro is the annual scheduling risk calculation formula of the aggregator. Through the optimization operation, the energy storage capacity Q is optimized:

[0237] Q=Q u +Q l (29)

[0238] Considering that the energy storage power configuration will affect its climbing rate, the constraint function of the energy storage power P is set:

[0239] P ≥ E t ×λ2 (30)

[0240] The calculation method of the main parameters is as follows:

[0241] 1) Daily risk loss of electric vehicles participating in auxiliary services W d

[0242] After participating in the ancillary service market, electric vehicle companies can obtain ancillary service compensation. Through market scheduling, risks can be compensated, thereby reducing risk losses. This paper takes the North China ancillary service market as an example, and the calculation formula for risk losses is shown in (13).

[0243]

[0244]

[0245] Among them, F t K is the peak load auxiliary fee obtained by the market entity during period t; t is the market coefficient, which is the inverse of the average load factor of the thermal power units in the provincial grid. This paper temporarily sets it to 1.2; is the peak load contribution rate, E t is the peak load power in period t, is the benchmark peak load-shaving power in period t; t is the response period, which is 15 minutes in this paper; is the market clearing price in period t; W is the total daily revenue of ancillary services; T is the opening period of ancillary services.

[0246] 2) Daily risk cost of electric vehicle participation assistance

[0247] When electric vehicles participate in auxiliary services, if the difference between the actual valley filling capacity and the reported capacity exceeds the market regulations, there will be opportunity cost losses. The cost calculation method at a single point in time is shown in formula (15).

[0248]

[0249] Among them, A e_t is the penalty cost at time t; γ is the penalty coefficient; is the load deviation rate; λ1 is the penalty critical value.

[0250] The daily risk cost can be obtained by summing up the scheduling costs of 96 points.

[0251] 3) Risk loss of energy storage directly involved in ancillary services

[0252] Among them, E b_battery is the total capacity of the energy storage configuration; the superscript d represents the parameter conditions on the dth day; The ancillary service risk loss of the energy storage system directly participating in the ancillary service at time t; The risk loss of peak and valley electricity prices obtained when energy storage resources are discharged during non-market periods; Total risk of loss for energy storage participating in the ancillary services market; The risk loss of coordinating the energy storage of electric vehicles to participate in ancillary services during period t; To coordinate the peak and valley electricity price risk losses of energy storage participating in ancillary services; To coordinate energy storage to assist electric vehicles and reduce the opportunity risk loss; W l d To coordinate the total risk loss of energy storage.

[0253] Among them, the calculation formula for direct benefits of ancillary services is shown in formula (16); the benefits of ancillary service electricity prices are shown in formula (17); and the total risk loss of the ancillary service market is shown in formula (18).

[0254]

[0255]

[0256]

[0257] Among them, P battery_u_t is the charging power of the energy storage participating in the ancillary service market at time t; p p and p v They are the energy storage discharge price and charging price respectively; Energy storage capacity to participate in the ancillary services market.

[0258] 4) Coordinate the energy storage risk loss W of electric vehicles l d

[0259] The revenue of the ancillary service market of coordinated energy storage is shown in formula (19); the peak-valley electricity price difference revenue of coordinated energy storage is shown in formula (20); the opportunity loss calculation reduced by coordinated energy storage is shown in formula (21); and the total revenue of coordinated energy storage is shown in formula (22).

[0260]

[0261]

[0262]

[0263]

[0264] Among them, P battery_l_t To coordinate the charging power of electric vehicles’ energy storage participating in the auxiliary service market at time t; Q l Energy storage capacity to coordinate electric vehicles to participate in auxiliary services; E t_b is the capacity of energy storage auxiliary regulation at time t.

[0265] 5) Annual cost of energy storage configuration A b

[0266] In the calculation of energy storage configuration cost, the annual value calculation formula of battery energy storage system is used as the energy storage cost calculation method. The calculation method of relevant parameters is as follows:

[0267]

[0268]

[0269] α=λ·β (25)

[0270]

[0271] L eye,D =ηD θ (27)

[0272] Among them, A b is the annual cost of the energy storage system; P bess is the planned energy storage power value; α is the energy storage amortization power cost; E bess is the planned value of energy storage capacity; β is the amortization capacity cost of energy storage; i is the interest rate level; T life is the energy storage service life; C E is the investment cost of the energy storage system; C m is the maintenance cost of the energy storage system; λ is the ratio of power investment to capacity investment of the energy storage system; K eye,D(i) The depth of charge and discharge is D (i) Battery life of 100 hours; L eye,D is the battery life of the lithium battery when the charge and discharge depth is D; η and θ are important parameters of the charge and discharge life of the lithium battery, which need to be determined based on the experimental fitting results. In this case, η is set to 694 and θ is set to -0.795.

[0273] Step 6: Use the particle swarm algorithm to solve the model, and the output result is the configuration result of the optimal energy storage capacity and energy storage power.

[0274] The error range of the ancillary service market is set to no more than 5%, and the ancillary service clearing price is 300 yuan / MWh. The peak-valley price difference is 1,030 yuan / MWh and the peak-flat price difference is 500 yuan / MWh based on the general industrial and commercial electricity price in Beijing. The investment cost is 5,166,000 yuan / MWh, the maintenance cost is 174,000 yuan / MWh / year

[28] , the ratio of power investment to capacity investment cost of the λ energy storage system is 1.172, and the market interest rate level is 8%. Based on this, the following simulation results are obtained.

[0275] Based on the optimal energy storage configuration strategy, the reported capacity, electric vehicle load proportion, energy storage load proportion, coordinated scheduling capacity and opportunity loss are calculated. The calculation results are shown in Table 2.

[0276] Table 2 Typical day market operation simulation results:

[0277]

[0278] During daily operation, energy storage is adjusted through continuous charging and discharging to make the load curve of electric vehicles as close as possible to the actual reported curve, thereby maximizing the market benefits of electric vehicles participating in auxiliary services.

[0279] It should be emphasized that the embodiments described in the present invention are illustrative rather than restrictive. Therefore, the present invention includes but is not limited to the embodiments described in the specific implementation manner. Any other implementation manners derived by those skilled in the art based on the technical solution of the present invention also fall within the scope of protection of the present invention.

Claims

1. A method for optimizing energy storage configuration to coordinate electric vehicles to participate in ancillary service markets, characterized in that: The following steps are involved: Step 1: Collect the historical load of electric vehicles and use the least squares support vector machine to predict the daily 96-point reported capacity of electric vehicles participating in the auxiliary service market; Step 2: The daily total load distribution of electric vehicles is obtained by fitting the historical total load distribution of electric vehicles collected in step 1; Step 3: Subtract the daily electric vehicle capacity predicted in step 1 from the actual load value to obtain the response error, and determine the error distribution between the reported capacity and the actual response capacity; based on the daily total load distribution function of electric vehicles obtained in step 2, use Monte Carlo to simulate the response capacity scale of future electric vehicle loads; Step 4: Based on the simulation results of the response capacity scale of the future electric vehicle load obtained in step 3, the energy storage demand capacity for coordinating the electric vehicles to participate in the auxiliary services on a daily basis is calculated through CVaR; Step 5: Based on the daily 96-point reported capacity curve obtained in step 1 and the daily electric vehicle response capacity scale obtained in step 3, combined with the supplementary difference capacity under the energy storage resource capacity configuration in step 4, construct an objective function to minimize the risk of auxiliary service scheduling, consider the risk loss of different response errors, and achieve optimal configuration of energy storage capacity; Step 6: Solve the energy storage model of step 5 using a particle swarm algorithm, and output the configuration results of the optimal energy storage capacity and energy storage power; The step 1 collects the historical load conditions of electric vehicles, including the annual daily load values ​​and the daily 96-point load values; The specific steps of using the least squares support vector machine to predict the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market in step 1 include: (1) Establish a nonlinear mapping The optimal decision function for the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is: Where w∈R k (k>d) is a high-dimensional feature weight vector, which means the influence of each key influencing factor on the 96-point load value of a single electric vehicle. i ∈R d is the training sample input of 96-point load of d-dimensional electric vehicle, b∈R is the bias; (2) According to the principle of structural risk minimization, the optimal w and b are found, and the optimization problem of predicting the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is obtained as follows: Among them, r>0 is the penalty parameter, ξ i is a slack variable, which is a constant that affects the prediction accuracy and can be solved by the PSO algorithm; Compared with the standard support vector machine algorithm, there are differences in constraints, as shown in Formula 3; y i ∈R is the training sample output of 96-point load of electric vehicle; (3) Applying the Lagrangian function to solve the optimization problem, we get: Among them, α i is the Lagrange multiplier vector, and α i >0, i=1,2,...,n; The final function for predicting the daily 96-point reporting capacity of electric vehicles participating in the auxiliary service market is shown in formula (5): in, The kernel function satisfies the Mercer condition. The radial basis kernel function is used as the kernel function. The kernel function formula is shown in formula (6): Where g is the width coefficient of the kernel function, which is a constant; The specific method of step 2 is: In the process of curve fitting, the annual electric vehicle charging load value data is collected, and samples are selected by random sampling. The samples are fitted with normal distribution, where the normal distribution function is as shown in formula (7): Then, the partial derivative is taken to obtain the parameter estimation values ​​as shown in formula (8) and formula (9): Among them, μ and σ are the sample mean and variance of the daily total load of electric vehicles, which are important parameters of the normal distribution function; and are the estimated values ​​of the two parameters μ and σ respectively; N is the number of daily total load samples of electric vehicles; x i is the load value of the ith electric vehicle daily total load sample; X is the electric vehicle daily total load sample set; The specific steps of step 3 include: (1) According to the least squares support vector machine prediction method proposed in step 1, the capacity plan reported by electric vehicles participating in the auxiliary service market on the next day is obtained by multiple cycles, and the actual response load value is subtracted to obtain the actual response error, and the error distribution is statistically analyzed; (2) fitting the daily total load distribution function of electric vehicles obtained in step 2, and simulating the response capacity scale of electric vehicle loads for 365 days a year through Monte Carlo simulation; Among them, the functional form of the fitting function is as follows: Among them, u is a random number that obeys the uniform distribution of [0,1], is the probability of the daily total load of electric vehicles being distributed in each stage, and z is the response capacity value of the daily total load of electric vehicles that obeys the normal distribution; The specific method of step 4 is: The CVaR difference capacity calculation method based on a certain confidence level can be expressed as formula (13). Under this energy storage capacity agreement, the electric vehicle response capacity is guaranteed to be supplemented at a certain confidence level: P(f(t,P)≤VaR(p))=1-p (12) Among them, f(t,P) is the shortfall power of the electric vehicle; P t is the reported capacity value of the electric vehicle at time t; p is the confidence level; E(P) is the expected power value of the electric vehicle at time t; is the energy storage capacity of the daily coordinated electric vehicle participation in the auxiliary service market response estimated by CVaR, m is the number of data taken; a is the VaR value; β is the confidence level; The specific steps of step 5 include: (1) In the process of optimizing the energy storage configuration for coordinating electric vehicles to participate in the ancillary service market, in order to minimize the annual dispatch risk of aggregators in the ancillary service market, the objective function of the energy storage optimization configuration model is established (14): Among them, Pro is the annual scheduling risk calculation formula of the aggregator, W d The daily risk loss of electric vehicles participating in auxiliary services on day d; is the daily risk cost of electric vehicle assistance at time t on day d; is the risk loss of energy storage directly participating in ancillary services on day d, Coordinate the energy storage risk loss of electric vehicles for day d; A b The annual cost of energy storage configuration; through optimization operation, optimize the energy storage capacity Q: Q=Q u +Q l (15) (2) Considering that the energy storage power configuration will affect its climbing rate, the constraint function of the energy storage power P is set: P≥E t ×λ2 (16) E t is the total charging power of the electric vehicle at time t; λ2 is the critical proportion of the charging power of the electric vehicle that does not participate in the valley filling auxiliary service; The calculation method of the parameters is as follows: ① Daily risk loss of electric vehicles participating in auxiliary services W d After participating in the ancillary service market, electric vehicle companies can obtain ancillary service compensation and compensate for risks through market scheduling, thereby reducing risk losses. The calculation formula for risk losses is shown in (17): Among them, F t K is the peak load auxiliary fee obtained by the market entity during period t; t is the market coefficient, which is the inverse of the average load factor of the thermal power units in the provincial grid. is the peak load contribution rate, E t is the peak load power in period t, is the reference peak load power in period t; t is the response period, is the market clearing price in period t; W is the total daily revenue of auxiliary services; T is the opening period of auxiliary services; ② Daily risk cost of electric vehicle assistance When electric vehicles participate in auxiliary services, if the difference between the actual valley filling capacity and the reported capacity exceeds the market regulations, there will be opportunity cost losses. The cost calculation method at a single point in time is as shown in formula (19): Among them, A e_t is the penalty cost at time t; γ is the penalty coefficient; is the load deviation rate; λ1 is the penalty critical value; The daily risk cost can be obtained by summing up the scheduling costs of 96 points. ③ Risk loss of energy storage directly involved in ancillary services The calculation formula of direct benefit of ancillary services is shown in formula (20); the benefit of ancillary service electricity price is shown in formula (21); the total risk loss of ancillary service market is shown in formula (22): Among them, P battery_u_t is the charging power of the energy storage participating in the ancillary service market at time t; p p and p v They are the energy storage discharge price and charging price respectively; Energy storage capacity to participate in the ancillary services market; Among them, E b_battery is the total capacity of the energy storage configuration; the superscript d represents the parameter conditions on the dth day; The ancillary service risk loss of the energy storage system directly participating in the ancillary service at time t; The risk loss of peak and valley electricity prices obtained when energy storage resources are discharged during non-market periods; Total risk of loss for energy storage participating in the ancillary services market; The risk loss of coordinating the energy storage of electric vehicles to participate in ancillary services during period t; To coordinate the peak and valley electricity price risk losses of energy storage participating in ancillary services; To coordinate the reduced opportunity risk loss after energy storage assists electric vehicles; To coordinate the total risk loss of energy storage; ④Coordinate the energy storage risk loss of electric vehicles The revenue of the ancillary service market of coordinated energy storage is shown in formula (23); the peak-valley electricity price difference revenue of coordinated energy storage is shown in formula (24); the opportunity loss calculation reduced by coordinated energy storage is shown in formula (25); the total revenue of coordinated energy storage is shown in formula (26): Among them, P battery_l_t To coordinate the charging power of electric vehicles’ energy storage participating in the auxiliary service market at time t; Q l Energy storage capacity to coordinate electric vehicles to participate in auxiliary services; E t_b is the capacity of energy storage auxiliary regulation at time t; ⑤Annual cost of energy storage configuration A b In the calculation of energy storage configuration cost, the annual value calculation formula of battery energy storage system is used as the energy storage cost calculation method. The calculation method of relevant parameters is as follows: α=λ×β (29) L eye,D =ηD θ (31) Among them, A b is the annual cost of the energy storage system; P bess is the planned energy storage power value; α is the energy storage amortization power cost; E bess is the planned value of energy storage capacity; β is the amortization capacity cost of energy storage; i is the interest rate level; T life is the energy storage service life; C E is the investment cost of the energy storage system; C m is the maintenance cost of the energy storage system; λ is the ratio of power investment to capacity investment of the energy storage system; K eye,D(i) The depth of charge and discharge is D (i) Battery life of 100 hours; L eye,D is the battery life of the lithium battery when the charge and discharge depth is D; η and θ are important parameters of the charge and discharge life of the lithium battery, which need to be determined based on the experimental fitting results.

2. The energy storage optimization configuration method for coordinating electric vehicles to participate in the auxiliary service market according to claim 1, characterized in that: The calculation steps of the particle swarm algorithm in step 6 include: (1) Parameter determination: Initialize the particle swarm algorithm parameters, including population size, initial position, and initial velocity; The initial position X0 and initial speed V0 are randomly generated, and the population size is set to 5; the initial position represents the initial energy storage capacity and power configuration, and the initial speed is the power change step, which is set not to exceed 1; (2) Fitness value calculation: Calculate the fitness value of each particle and update the position and velocity of the particle; the fitness value is the objective function in step 5, and the calculation formula is as follows (14); the velocity update formula is formula (32): V k+1 = w×V k + c1 × rand()× (p best - X k )+ c2 ×rand()× (g best -X k ) (32) Among them, V k+1 is the updated speed; w is the inertia factor; V k is the current speed; rand() is a random number generation function that generates a random number between 0 and 1; p best is the optimal position currently achieved by the particle; X k is the current position of the particle; c1 and c2 are learning factors, which learn from the best historical position of the particle and the best current position in the population respectively; The position update formula is formula (33): X k+1 =X k + V k+1 (33) Finally, the optimal energy storage capacity and power configuration can be obtained through optimization results.