Electric vehicle participation power grid frequency modulation auxiliary service incentive pricing method

Abstract

The invention relates to the technical field of power distribution networks, and particularly discloses an incentive pricing method for participation of an electric vehicle in a power grid frequency modulation auxiliary service, which comprises the following steps of: establishing a user characteristic index system aiming at behavior characteristics in a process of participation of a user in a secondary frequency modulation service, and quantifying multi-dimensional characteristics of the user; establishing a sample matrix according to the index quantized value, and predicting the number of frequency modulation participation time periods of the user based on a multi-layer perceptron; the frequency modulation compensation income of the aggregator is evaluated, and corresponding frequency modulation compensation excitation is distributed to the user based on a Shapley value; in addition, the user incentive further comprises charging income brought to the aggregator by the user participating in the frequency modulation service, and personalized incentive based on the historical performance of the user. The method helps to improve the enthusiasm of the user to participate in the secondary frequency modulation auxiliary service, guides the user to be off-network within the predicted time, reduces the uncertainty of the aggregator in the regulation and control of the electric vehicle, and achieves the win-win situation of the aggregator and the user.

Description

Technical Field

[0001] The present invention belongs to the technical field of power distribution networks, and in particular relates to an incentive pricing method for electric vehicles to participate in power grid frequency regulation auxiliary services. Background Art

[0002] All regions of my country are actively promoting the construction of the frequency regulation auxiliary service market, and are committed to mobilizing the enthusiasm of flexible resources to participate in system regulation. Electric vehicles, with their advantages of fast response and flexible adjustment, are considered to have great potential in participating in secondary frequency regulation auxiliary services. Currently, large-scale vehicle-grid interaction pilot projects have been carried out under the guidance of the government.

[0003] As an intermediary connecting electric vehicles and the electricity market, electric vehicle aggregators obtain economic benefits by aggregating and regulating scattered electric vehicle resources to participate in secondary frequency regulation auxiliary services. In the process of aggregators coordinating and regulating electric vehicles to participate in grid auxiliary services, incentive-driven methods are usually used to guide users to voluntarily accept changes in the charging status of electric vehicles, so as to accurately respond to frequency regulation instructions and increase users' enthusiasm for participating in services. At present, a large number of studies have reported on incentive pricing strategies to guide electric vehicle users to participate in demand response, but they mainly focus on achieving user incentive pricing through a single factor, without considering the multi-dimensional portrait characteristics of users during the response period, and there is a relative lack of research on the issue of incentive pricing for electric vehicle users to participate in secondary frequency regulation auxiliary services. Summary of the invention

[0004] The purpose of the present invention is to provide an incentive pricing method for electric vehicles to participate in power grid frequency regulation auxiliary services. By quantifying the multi-dimensional portrait characteristics of users during the service period, personalized incentive pricing is provided before the user officially charges, aiming to guide users to participate in secondary frequency regulation auxiliary services fairly and orderly, and to leave the grid within the expected time.

[0005] To achieve the above-mentioned purpose, the present invention provides a method for pricing incentives for electric vehicles to participate in power grid frequency regulation auxiliary services, comprising:

[0006] In the user feature portrait modeling stage, based on user historical data, 9 feature indicators in different dimensions are extracted for the scenario of electric vehicle users participating in secondary frequency regulation auxiliary services, thereby constructing a multi-dimensional user feature indicator system.

[0007] In the stage of predicting the number of FM time periods in which target users participate, the quantitative value of user characteristic indicators is used as the model input, and the number of FM time periods in which users actually participate is used as the model output, forming the sample matrix of MLP (Multi-Layer Perceptron, MLP) prediction, which is used to predict the number of FM time periods in which target users are expected to participate.

[0008] In the target user incentive pricing stage, user incentives consist of charging incentives, frequency modulation compensation incentives and personalized incentives. Among them, the charging incentive is determined by the user's expected participation time; the frequency modulation compensation incentive is obtained by using the Shapley value to fairly and reasonably distribute the frequency modulation compensation income of the aggregator; the personalized incentive is obtained by evaluating whether the user's past participation performance is outstanding, and giving additional incentive fees accordingly.

[0009] Therefore, before users officially start charging, aggregators can effectively guide users to participate in secondary frequency regulation auxiliary services in an orderly manner by providing corresponding incentive fees and expected off-grid time.

[0010] Preferably, a user characteristic index system is established, and the 9 characteristic indexes and their definition methods are as follows:

[0011] 1) Charging power: In the process of aggregators coordinating and regulating electric vehicles, the limit of electric vehicle charging power determines the adjustable margin of power, which defines the calculation method of charging power index:

[0012] I CP =P max -P min

[0013] Where: P max , P min They are the maximum and minimum charging power of electric vehicles determined by the charging equipment.

[0014] 2) Charging time: The survey results of NHTS2017 show that about 70% of users are accustomed to charging their electric vehicles only when they get home at night. Therefore, the charging time indicator is defined as the charging time of electric vehicles in disordered charging scenarios:

[0015]

[0016] Where: η 100 P is the power consumption of electric vehicles per 100 kilometers; N is the rated charging power of the electric vehicle. To give full play to the adjustable potential of the electric vehicle, P start =I P / 2; d is the daily driving distance of the electric vehicle.

[0017] 3) Charging capacity: The adjustable margin of EV power is determined by the initial SOC, target SOC, and battery capacity. The charging capacity index is obtained by the following formula:

[0018] I CE =C EV (SOC aim -SOC start )

[0019] Where: CEV is the battery capacity of electric vehicles; SOC aim The target SOC set by the user based on the prompt of the smart charging pile; SOC start is the SOC of the electric vehicle before charging begins, and its expression is:

[0020]

[0021] Where: SOC end It is the SOC of the electric vehicle at the end of the previous charging process.

[0022] 4) Regulation rate: The regulation rate index refers to the power change of the electric vehicle per unit time, that is, the rate at which the electric vehicle responds to the frequency modulation command:

[0023]

[0024] 5) Response time: The response time index refers to the difference between the time when the electric vehicle receives the frequency modulation command and the time when it responds to the frequency modulation command:

[0025]

[0026] 6) Regulation accuracy: The regulation accuracy index refers to the deviation between the actual output value and the command value after the electric vehicle responds to the frequency modulation command:

[0027]

[0028] The allowable error of power generation unit regulation is 1.5% of its rated output.

[0029] 7) Bid Winning: The aggregator’s bid winning of frequency modulation capacity during the target user’s charging period will determine the aggregator’s demand for EV frequency modulation response. Therefore, this indicator is defined by estimating the average bid winning during the user’s charging period:

[0030]

[0031] Where: c(t i ) is t i The winning capacity of the aggregator in the frequency modulation period is obtained from the pre-arrangement results announced by the frequency modulation market a few days ago; M un The number of frequency regulation periods that electric vehicles are expected to participate in under the scenario of disorderly charging can be calculated as:

[0032]

[0033] 8) Response willingness: Given the uncertainty of users’ response to frequency modulation instructions, the response willingness index is defined as the historical response rate of users in the past 30 days, and its expression is:

[0034]

[0035] Where: J is the number of responses initiated by nearly 30 aggregators to electric vehicle users through smart charging pile terminals; x j =1 and x j =0 respectively represent the user's choice of responding or not responding to the frequency modulation instruction at the terminal when initiating a response for the jth time.

[0036] 9) Performance: Whether the user leaves the grid at the agreed time is related to the charging income and frequency regulation compensation income of the aggregator after the charging is completed, and users with good historical performance often have a greater probability of fulfilling the contract in this response. The calculation method of the performance indicator is defined as follows:

[0037]

[0038] Where: K is the number of times the electric vehicle has been charged in the past 30 days; the actual charging time refers to the time when the charging current is not zero; the charging pile occupation time refers to the difference between the time when the electric vehicle is connected to the grid and when it is off the grid.

[0039] Preferably, the quantitative values ​​of seven characteristic indicators, namely, charging power, charging time, charging power, regulation rate, response time, regulation accuracy, and aggregator bidding situation, are selected as the input samples of the MLP model, and the actual frequency modulation period number of the sample user i participating in the secondary frequency modulation service is is the output sample of the MLP model, and the model sample matrix Q consisting of m sample users is obtained:

[0040]

[0041] Preferably, in the above technical solution, the following steps are used to train and verify the MLP model and perform prediction:

[0042] 1) Data preprocessing: Create a sample matrix Q and divide it into a training set and a validation set in a ratio of 7:3;

[0043] 2) Model initialization: Each layer of neurons in the MLP model is configured as fully connected. Specifically, the model contains a 7-neuron input layer, which is responsible for accepting the quantized values ​​of user feature indicators in 7 dimensions as input; followed by two hidden layers, each with 5 neurons, using the Sigmoid activation function to add nonlinear transformation to facilitate feature extraction; finally, the output layer is configured as a single neuron, which directly outputs the predicted results of the number of users participating in the FM period.

[0044] 3) Training model: The SGD optimizer was used to train the model for 150 rounds. During the weight update process, the L2 regularization strategy was used to limit the size of the weights, thereby enhancing the generalization ability of the model.

[0045] 4) Validation model: Use the validation set to evaluate the effect of the trained model, select the mean square error as the loss function to calculate the loss value on the validation set, and save the model with the minimum validation loss as the optimal model;

[0046] 5) Execute prediction: load the previously saved optimal model, input the quantitative value of the target user characteristic index into the model, and the model directly outputs the prediction result of the number of target users participating in the frequency modulation period

[0047] Preferably, in the above technical solution, the incentive fee that the target user can obtain by participating in the secondary frequency modulation auxiliary service is as follows:

[0048]

[0049] Where: α e The sharing factor of the charging revenue that the aggregator gives to the user; The charging revenue brought to the aggregator by the target user i participating in the secondary frequency regulation auxiliary service during the whole period; α m The coefficient of the aggregator's share of the frequency modulation compensation revenue to users. The frequency modulation compensation income brought to the aggregator by the target user i participating in the secondary frequency modulation auxiliary service at all times; Provide personalized incentives for target user i to participate in secondary frequency regulation auxiliary services.

[0050] Preferably, in the above technical solution, the charging income in the user incentive composition The calculation method is as follows:

[0051]

[0052] Where: i =1,υ i =0,υ i =-1 respectively indicates that user i's participation in secondary frequency regulation auxiliary services will increase, remain unchanged, or reduce the charging revenue of the aggregator; DM i It represents the absolute value of the difference in the number of frequency regulation periods that user i participates in under the scenarios of disordered charging and participating in secondary frequency regulation auxiliary service; is the charging fee corresponding to the frequency modulation period t.

[0053] Preferably, in the above technical solution, the frequency modulation compensation in the user incentive composition The calculation method refers to the allocation method based on Shapley value. Specifically, assuming that the set of all users connected to the aggregator during the frequency modulation period t is N, and S is any alliance of N that includes user i, then the income that user i can obtain from participating in the secondary frequency modulation auxiliary service is:

[0054]

[0055] Where: |N| is the total number of grid-connected users; |S| is the number of users in alliance S; v(S) is the frequency regulation compensation income of each sub-alliance scenario, which depends on the frequency regulation compensation and its default penalty, and is defined as:

[0056]

[0057] Preferably, in the above technical solution, the frequency regulation compensation consists of two parts: frequency regulation mileage compensation and frequency regulation capacity compensation. The frequency regulation mileage compensation reflects the actual regulation effect of the alliance S, which is determined by the frequency regulation performance and regulation mileage:

[0058]

[0059] Where: Compensation for the frequency modulation mileage of alliance S during frequency modulation period t; K is the unit frequency modulation mileage price in frequency modulation period t; t (S) is the comprehensive frequency regulation performance index of alliance S in frequency regulation period t, according to the regulation rate of |S| electric vehicles Response time Adjustment accuracy The weighted average of t (S) is the actual FM mileage provided by alliance S during FM period t, defined as the overall power change compared to FM period t-1.

[0060] The frequency regulation capacity compensation is determined by the frequency regulation capacity that won the bid the day before, and is compensated according to the unified frequency regulation capacity price:

[0061]

[0062] Where: is the frequency regulation capacity compensation of alliance S in frequency regulation period t; R C is the price per unit frequency modulation capacity; C t is the FM capacity requisitioned by the aggregator in FM period t.

[0063] During the period when Alliance S provided secondary frequency regulation auxiliary services, the comprehensive frequency regulation performance index K was not achieved. t (S) Aggregators with corresponding thresholds will be deducted v for the corresponding period M (S) and Using 0-1 variables Indicates whether the relevant compensation is deducted:

[0064]

[0065] Where: Indicates that alliance S is not deducted v during the FM period t M (S), otherwise Indicates that alliance S is not deducted v during the FM period t C (S), otherwise

[0066] Preferably, in the above technical solution, the personalized incentive in the user incentive composition The calculation method is as follows: For users who have a good historical performance in participating in the secondary frequency modulation auxiliary service, personalized incentives are given to them based on whether the average quantitative values ​​of their response willingness index and performance index reach the incentive threshold set by the aggregator:

[0067]

[0068] Where: α f is the personalized incentive coefficient; σ h The threshold for users to obtain incentives; is the average quantitative value of the response willingness index and the fulfillment performance index of user i, and the calculation method is:

[0069]

[0070] Compared with the existing technology, the present invention has the following beneficial effects:

[0071] 1) Aiming at the scenario where electric vehicle users participate in secondary frequency regulation auxiliary services, an innovative multi-dimensional user characteristic indicator system is constructed, which not only includes the adjustment ability and charging preference of users when responding, but also takes into account the frequency regulation needs of aggregators, forming a comprehensive and non-redundant user characteristic portrait model.

[0072] 2) To solve the uncertainty of user off-grid time, a user historical feature sample matrix was constructed based on the proposed user feature quantification method, and the MLP model was trained using the matrix. The model can accurately and quickly predict the number of frequency modulation time periods that target users are expected to participate in, and on the basis of providing time support for incentive pricing, guide users to complete charging within the specified time and leave the grid on time.

[0073] 3) User incentive pricing consists of charging revenue, frequency regulation compensation and personalized incentives. Among them, charging revenue and frequency regulation compensation are allocated to user incentives based on the user's contribution; while personalized incentives are aimed at increasing user participation enthusiasm and giving additional incentive fees based on the user's subjective performance. Therefore, under the premise of ensuring that the aggregator has considerable profits, user incentive pricing is set according to user response capabilities, guiding users to actively participate in secondary frequency regulation auxiliary services, and ultimately achieving a mutually beneficial and win-win situation for both parties. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only one embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0075] Figure 1 It is a flow chart of a method for pricing incentives for electric vehicles to participate in power grid frequency regulation auxiliary services of the present invention;

[0076] Figure 2 It is one of the characteristic index correlation coefficient diagrams of the present invention. DETAILED DESCRIPTION

[0077] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0078] In the description of the present invention, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed" etc. are understood to exclude the number itself, and "above", "below", "within" etc. are understood to include the number itself. If the terms "first", "second", "third" are described, they are only used for the purpose of description and distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0079] like Figure 1 As shown in the figure, the incentive pricing method for electric vehicles to participate in power grid frequency regulation auxiliary services includes the following steps:

[0080] Step 1: Obtain the historical charging data of electric vehicle users and use it as a sample user data set;

[0081] Step 2: According to the definition method of the 9 characteristic indicators in the characteristic indicator system, the characteristic indicators of all sample users are quantified according to the sample user data set, and a characteristic portrait model of users participating in the secondary frequency modulation auxiliary service is established;

[0082] Step 3: The quantified values ​​of the seven characteristic indicators of the sample users, namely, charging power, charging time, charging power, regulation rate, response time, response accuracy, and bid winning situation, are used as the input of the model, and the actual number of frequency modulation periods participated by the users is used as the output of the model, and the MLP neural network model is trained offline;

[0083] Step 4: For the target user, input the quantified value of the target user characteristic index of the same dimension into the trained MLP model, and the model will output the number of FM time slots that the target user is expected to participate in;

[0084] Step 5: Calculate the charging incentive in the user incentive based on the estimated charging revenue brought to the aggregator by the user's participation in the frequency regulation auxiliary service;

[0085] Step 6: Calculate the frequency modulation compensation incentive in the user incentive, and use the Shapley value to fairly and reasonably distribute the benefits based on the frequency modulation compensation available to the aggregator.

[0086] Step 7: Calculate the personalized incentives in the user incentives by evaluating whether the average quantitative values ​​of the user's response willingness and performance characteristic indicators reach the incentive threshold;

[0087] Step 8: Output the total incentive for users to participate in the secondary frequency modulation auxiliary service according to steps 5, 6 and 7;

[0088] Step 9: According to steps 4 and 8, the expected off-grid time and incentive fee of the user for participating in the secondary frequency regulation auxiliary service are obtained, so as to guide the user to actively participate in the service.

[0089] A method for incentivizing pricing of electric vehicles participating in power grid frequency regulation auxiliary services in this embodiment is described in more detail to enable those skilled in the art to better understand the present invention:

[0090] User feature model building stage: Figure 2 It shows the correlation coefficient between user feature indicators. It can be seen from the figure that indicator I CT With index I CE The correlation between them is strong because the indicator I CT The increase in the I means that the electric vehicle can adjust more power, which in turn corresponds to a longer charging time. CE It also further reflects the user's subjective preference for power-off during charging behavior. Except for the significant correlation between the above two indicators, the correlation coefficients between the other indicators are almost zero, which shows that the correlation between the characteristic indicators is weak and can accurately depict the user's distinctive portrait characteristics.

[0091] Prediction stage of the number of users participating in frequency modulation period: The SVR model and random forest model are introduced to compare with the MLP model to highlight the effectiveness of the MLP model in the scenario mentioned in this article. Table 1 shows the comparison table of frequency modulation period prediction models. It can be seen from Table 1 that the MLP model is superior to other models in terms of prediction accuracy and prediction efficiency, and can accurately and quickly predict the number of users participating in frequency modulation period prediction stage.

[0092] Table 1

[0093]

[0094] Target user incentive pricing stage: Assume that the aggregator sets α f =1.2,σ h is 0.7, and the aggregator's sharing coefficient for frequency modulation compensation and charging income is 0.5. Table 2 shows the incentives and their composition for the target user to participate in the frequency modulation service. It can be seen from Table 2 that the target user's final total incentive consists of frequency modulation compensation incentives, charging incentives, and personalized incentives based on the user's historical response performance. Among them, due to the high initial SOC of the target user and the short charging time, only the frequency modulation compensation incentives of four frequency modulation periods can be obtained. Therefore, the total frequency modulation compensation incentive is not much higher than the charging incentive. The quantitative value of the target user's charging preference label is calculated to be 0.73, which is higher than the threshold σ for obtaining personalized incentives. h , so users can get additional personalized incentives based on good historical performance.

[0095] Table 2

[0096]

[0097] In summary, the provided incentive pricing method for electric vehicles to participate in grid frequency regulation auxiliary services can provide incentive pricing for users fairly, accurately and quickly, help increase users' enthusiasm for participating in secondary frequency regulation auxiliary services, and at the same time guide users to leave the grid within the expected time, reducing the uncertainty of aggregators in regulating electric vehicles. This method achieves a win-win situation for aggregators and users.

[0098] The foregoing descriptions of specific exemplary embodiments of the present invention are for the purpose of explanation and illustration. These descriptions are not intended to limit the present invention to the precise form disclosed, and it is obvious that many changes and variations can be made based on the above teachings. Although an embodiment of the present invention has been shown and described, this specific embodiment is only an explanation of the present invention and is not a limitation of the invention. The specific features, structures, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present invention and its practical application, so that those skilled in the art can make modifications, substitutions, variations and various selections and changes to the embodiments without creative contribution as needed without departing from the principles and purpose of the present invention after reading this specification, but they are protected by patent law as long as they are within the scope of the claims of the present invention.

Claims

1. A method for incentivizing electric vehicles to participate in power grid frequency regulation auxiliary services, characterized in that: include: In the user feature portrait modeling stage, based on user historical data, 9 feature indicators in different dimensions are extracted for the scenario of electric vehicle users participating in secondary frequency regulation auxiliary services, thereby constructing a multi-dimensional user feature indicator system. In the stage of predicting the number of FM time periods in which target users participate, the quantitative value of user characteristic indicators is used as the model input, and the number of FM time periods in which users actually participate is used as the model output, forming the sample matrix of MLP (Multi-Layer Perceptron, MLP) prediction, which is used to predict the number of FM time periods in which target users are expected to participate. In the target user incentive pricing stage, user incentives consist of charging incentives, frequency modulation compensation incentives and personalized incentives. Among them, the charging incentive is determined by the user's expected participation time; the frequency modulation compensation incentive is obtained by using the Shapley value to fairly and reasonably distribute the frequency modulation compensation income of the aggregator; the personalized incentive is obtained by evaluating whether the user's past participation performance is outstanding and giving additional incentive fees accordingly. Therefore, before users officially start charging, aggregators can effectively guide users to participate in secondary frequency regulation auxiliary services in an orderly manner by providing corresponding incentive fees and expected off-grid time.

2. According to claim 1, an incentive pricing method for electric vehicles to participate in power grid frequency regulation auxiliary services is characterized by establishing a user characteristic index system, and the 9 characteristic indicators and their definition methods are as follows: 1) Charging power: In the process of aggregators coordinating and regulating electric vehicles, the limit of electric vehicle charging power determines the adjustable margin of power, which defines the calculation method of charging power index: I CP =P max -P min Where: P max , P min They are the maximum and minimum charging power of electric vehicles determined by the charging equipment. 2) Charging time: The survey results of NHTS2017 show that about 70% of users are accustomed to charging their electric vehicles only when they get home at night. Therefore, the charging time indicator is defined as the charging time of electric vehicles in the disorderly charging scenario: Where: η 100 P is the power consumption of electric vehicles per 100 kilometers; N is the rated charging power of the electric vehicle. To give full play to the adjustable potential of the electric vehicle, P start =I P / 2; d is the daily driving distance of the electric vehicle. 3) Charging capacity: The adjustable margin of EV power is determined by the initial SOC, target SOC, and battery capacity. The charging capacity index is obtained by the following formula: I CE =C EV (SOC aim -SOC start ) Where: C EV is the battery capacity of electric vehicles; SOC aim The target SOC set by the user based on the prompt of the smart charging pile; SOC start is the SOC of the electric vehicle before charging begins, and its expression is: Where: SOC end It is the SOC of the electric vehicle at the end of the previous charging process. 4) Regulation rate: The regulation rate index refers to the power change of the electric vehicle per unit time, that is, the rate at which the electric vehicle responds to the frequency modulation command: 5) Response time: The response time index refers to the difference between the time when the electric vehicle receives the frequency modulation command and the time when it responds to the frequency modulation command: 6) Regulation accuracy: The regulation accuracy index refers to the deviation between the actual output value and the command value after the electric vehicle responds to the frequency modulation command: in, The allowable error for power generation unit regulation is 1.5% of its rated output. 7) Bid Winning: The aggregator’s bid winning of frequency modulation capacity during the target user’s charging period will determine the aggregator’s demand for EV frequency modulation response. Therefore, this indicator is defined by estimating the average bid winning during the user’s charging period: Where: c(t i ) is t i The winning capacity of the aggregator in the frequency modulation period is obtained from the pre-arrangement results announced by the frequency modulation market a few days ago; M un The number of frequency regulation periods that electric vehicles are expected to participate in under the scenario of disorderly charging can be calculated as: 8) Response willingness: Given the uncertainty of users’ response to frequency modulation instructions, the response willingness index is defined as the historical response rate of users in the past 30 days, and its expression is: Where: J is the number of responses initiated by nearly 30 aggregators to electric vehicle users through smart charging pile terminals; x j =1 and x j =0 respectively represent the user's choice of responding or not responding to the frequency modulation instruction at the terminal when initiating a response for the jth time. 9) Performance: Whether the user leaves the grid at the agreed time is related to the charging income and frequency regulation compensation income of the aggregator after the charging is completed, and users with good historical performance often have a greater probability of fulfilling the contract in this response. The calculation method of the performance indicator is defined as follows: Where: K is the number of times the electric vehicle has been charged in the past 30 days; the actual charging time refers to the time when the charging current is not zero; the charging pile occupation time refers to the difference between the time when the electric vehicle is connected to the grid and when it is off the grid.

3. The method for incentivizing electric vehicles to participate in power grid frequency regulation auxiliary services according to claim 1, characterized in that: The quantitative values ​​of seven characteristic indicators, including charging power, charging time, charging power, regulation rate, response time, regulation accuracy, and aggregator bidding situation, are selected as the input samples of the MLP model. The actual number of frequency modulation periods in which sample user i participates in the secondary frequency modulation service is the output sample of the MLP model, and the model sample matrix Q consisting of m sample users is obtained:

4. The method for incentivizing electric vehicles to participate in power grid frequency regulation auxiliary services according to claim 3, characterized in that: The following steps are used to train, validate the MLP model, and perform predictions: 1) Data preprocessing: Create a sample matrix Q and divide it into a training set and a validation set in a ratio of 7:3; 2) Model initialization: Each layer of neurons in the MLP model is configured as fully connected. Specifically, the model contains a 7-neuron input layer, which is responsible for accepting the quantized values ​​of user feature indicators in 7 dimensions as input; followed by two hidden layers, each with 5 neurons, using the Sigmoid activation function to add nonlinear transformation to facilitate feature extraction; finally, the output layer is configured as a single neuron, which directly outputs the predicted results of the number of users participating in the FM period. 3) Training model: The SGD optimizer was used to train the model for 150 rounds. During the weight update process, the L2 regularization strategy was used to limit the size of the weights, thereby enhancing the generalization ability of the model. 4) Validation model: Use the validation set to evaluate the effect of the trained model, select the mean square error as the loss function to calculate the loss value on the validation set, and save the model with the minimum validation loss as the optimal model; 5) Execute prediction: load the previously saved optimal model, input the quantitative value of the target user characteristic index into the model, and the model directly outputs the prediction result of the number of target users participating in the frequency modulation period 5. The method for incentivizing electric vehicles to participate in power grid frequency regulation auxiliary services according to claim 1, characterized in that: The incentive fees available to target users for participating in secondary frequency regulation auxiliary services are as follows: Where: α e The sharing factor of the charging revenue that the aggregator gives to the user; The charging revenue brought to the aggregator by the target user i participating in the secondary frequency regulation auxiliary service during the whole period; α m The coefficient of the aggregator's share of the frequency modulation compensation revenue to users. The frequency modulation compensation income brought to the aggregator by the target user i participating in the secondary frequency modulation auxiliary service at all times; Provide personalized incentives for target user i to participate in secondary frequency regulation auxiliary services.

6. The method for incentivizing electric vehicles to participate in power grid frequency regulation auxiliary services according to claim 5, characterized in that: Charging income in user incentive composition The calculation method is as follows: Where: i =1,υ i =0,υ i =-1 respectively indicates that user i's participation in secondary frequency regulation auxiliary services will increase, remain unchanged, or reduce the charging revenue of the aggregator; DM i It represents the absolute value of the difference in the number of frequency regulation periods that user i participates in under the scenarios of disordered charging and participating in secondary frequency regulation auxiliary service; is the charging fee corresponding to the frequency modulation period t.

7. The method for incentivizing electric vehicles to participate in power grid frequency regulation auxiliary services according to claim 5, characterized in that: Frequency Modulation Compensation in User Incentive Composition The calculation method refers to the allocation method based on Shapley value. Specifically, assuming that the set of all users connected to the aggregator during the frequency modulation period t is N, and S is any alliance of N that includes user i, then the income that user i can obtain by participating in the secondary frequency modulation auxiliary service is: Where: |N| is the total number of grid-connected users; |S| is the number of users in alliance S; v(S) is the frequency regulation compensation income of each sub-alliance scenario, which depends on the frequency regulation compensation and its default penalty, and is defined as:

8. The method for incentivizing electric vehicles to participate in power grid frequency regulation auxiliary services according to claim 7, characterized in that: Frequency regulation compensation consists of two parts: frequency regulation mileage compensation and frequency regulation capacity compensation. Frequency regulation mileage compensation reflects the actual regulation effect of Alliance S, which is determined by frequency regulation performance and regulation mileage: Where: Compensation for the frequency modulation mileage of alliance S during frequency modulation period t; K is the unit frequency modulation mileage price in frequency modulation period t; t (S) is the comprehensive frequency regulation performance index of alliance S in frequency regulation period t, according to the regulation rate of |S| electric vehicles Response time Adjustment accuracy The weighted average of t (S) is the actual FM mileage provided by alliance S during FM period t, defined as the overall power change compared to FM period t-1. The frequency regulation capacity compensation is determined by the frequency regulation capacity that won the bid the day before, and is compensated according to the unified frequency regulation capacity price: Where: is the frequency regulation capacity compensation of alliance S in frequency regulation period t; R C is the price per unit frequency modulation capacity; C t is the FM capacity requisitioned by the aggregator in FM period t. During the period when Alliance S provided secondary frequency regulation auxiliary services, the comprehensive frequency regulation performance index K was not achieved. t (S) Aggregators with corresponding thresholds will be deducted v for the corresponding period M (S) and Using 0-1 variables Indicates whether the relevant compensation is deducted: Where: Indicates that alliance S is not deducted v during the FM period t M (S), otherwise Indicates that alliance S is not deducted v during the FM period t C (S), otherwise 9. The method for incentivizing electric vehicles to participate in power grid frequency regulation auxiliary services according to claim 5, characterized in that: Personalized incentives in user incentive composition The calculation method is as follows: For users who have a good historical performance in participating in the secondary frequency modulation auxiliary service, personalized incentives are given to them based on whether the average quantitative values ​​of their response willingness index and performance index reach the incentive threshold set by the aggregator: Where: α f is the personalized incentive coefficient; σ h The threshold for users to obtain incentives; is the average quantitative value of the response willingness index and the fulfillment performance index of user i, and the calculation method is:

Images