A day-ahead response capability evaluation method, a scheduling optimization method and equipment of an electric vehicle virtual power plant

Abstract

The application discloses a kind of day-ahead response capability evaluation method, scheduling optimization method and equipment of electric vehicle virtual power plant, and relates to the technical field of smart grid.The method is based on the idea of data-driven, using LSTM-GA combination method, captures the time-dependent relationship in the vast amount of historical operation data of virtual power plant, realizes the accurate identification of EV single non-fixed characteristic parameters, first, the EV group (i.e.the electric vehicle virtual power plant) is taken as the evaluation object, the day-ahead response capability of electric vehicle virtual power plant is evaluated, the accuracy of day-ahead response capability evaluation is improved, and the accuracy of optimized scheduling is further improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of smart grid, in particular to a day-ahead response capability evaluation method, a dispatch optimization method and equipment of an electric vehicle virtual power plant. BACKGROUND

[0002] Electric vehicles (EVs) have developed rapidly in recent years due to their energy-saving and zero-emission advantages. EV batteries have flexible bidirectional power regulation characteristics and can be controlled to change their charging and discharging states to participate in demand response of power systems. An electric vehicle virtual power plant (EVPP) is responsible for aggregating and managing EVs in a certain area, thereby providing a way for dispersed and small-capacity EVs to participate in optimal dispatch of power systems. A common way for the EVPP to participate in regulation is to respond to system dispatch requirements and provide power support. Accurate evaluation of the day-ahead response capability of the EVPP is crucial for the EVPP to serve power grids and EV users. Existing research generally gives a probability distribution of response characteristic parameters of EVs according to statistical experience, then uses the Monte Carlo method to perform random sampling, calculates the response capability of each EV generated by sampling, and adds them to obtain the final response capability of the EVPP. However, the probability distribution describes the general characteristics of the parameters and cannot truly reflect the differences between the parameters of the EV group in the actual scenario, making it difficult to accurately calculate the response capability. SUMMARY

[0003] The present application aims to provide a day-ahead response capability evaluation method, a dispatch optimization method and equipment of an electric vehicle virtual power plant to improve the accuracy of day-ahead response capability evaluation and thereby improve the accuracy of optimal dispatch.

[0004] To achieve the above-mentioned purpose, the present application provides the following solutions.

[0005] In a first aspect, the present application provides a day-ahead response capability evaluation method of an electric vehicle virtual power plant, comprising:

[0006] Based on historical operation data of each EV in the electric vehicle virtual power plant, the predicted value of a fixed characteristic parameter of each EV in the electric vehicle virtual power plant on the day to be predicted is obtained by using mean calculation, wherein the fixed characteristic parameter includes an allowed upper limit of the state of charge of the battery, an allowed lower limit of the state of charge of the battery and the capacity of the battery.

[0007] Based on historical operation data of each EV unit in the electric vehicle virtual power plant, a LSTM-GA combined method is used to obtain predicted values of non-fixed characteristic parameters of each EV unit in the electric vehicle virtual power plant on the day to be predicted, wherein the non-fixed characteristic parameters include: network charging state of life, minimum off-network power, network time and off-network time.

[0008] According to the predicted values of each characteristic parameter of each EV unit in the electric vehicle virtual power plant on the day to be predicted, each EV unit in the electric vehicle virtual power plant is clustered to obtain a plurality of clusters, wherein each characteristic parameter includes each fixed characteristic parameter and each non-fixed characteristic parameter.

[0009] The aggregation expression of each characteristic parameter in each cluster is calculated respectively.

[0010] According to the aggregation expression of each characteristic parameter in each cluster, the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant are calculated.

[0011] In a second aspect, the application provides a scheduling optimization method of an electric vehicle virtual power plant, comprising:

[0012] An optimization model is constructed with the goal of optimal tracking effect of actual response power on scheduling demand signals and minimum amount of abandoned wind and light.

[0013] Based on the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant, the constraint conditions of the optimization model are determined, wherein the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant are obtained by using the above-mentioned day-ahead response capability evaluation method of the electric vehicle virtual power plant.

[0014] Based on the constraint conditions, the optimization model is solved to obtain an optimal scheduling scheme.

[0015] In a third aspect, the application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned day-ahead response capability evaluation method of the electric vehicle virtual power plant, or the above-mentioned scheduling optimization method of the electric vehicle virtual power plant.

[0016] According to the specific embodiments provided by the application, the application has the following technical effects.

[0017] The application provides a day-ahead response capability evaluation method, a scheduling optimization method and equipment of an electric vehicle virtual power plant. The application is based on a data driving idea, adopts an LSTM-GA combined method, captures time dependence in massive historical operation data of the virtual power plant, and realizes accurate identification of non-fixed characteristic parameters of an EV unit. First, the EV group (i.e., the electric vehicle virtual power plant) is taken as an evaluation object to evaluate the day-ahead response capability of the electric vehicle virtual power plant, so that the accuracy of the day-ahead response capability evaluation is improved, and the accuracy of the optimized scheduling is further improved. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 A flowchart of a day-ahead response capability evaluation method of an electric vehicle virtual power plant is provided for an embodiment of the present application.

[0020] Figure 2 A schematic diagram of a maximum controllable area of an EV unit into a network is provided for an embodiment of the present application.

[0021] Figure 3 A relationship curve diagram of the number of clustering groups and SSE is provided for an embodiment of the present application.

[0022] Figure 4 An internal structure diagram of LSTM is provided for an embodiment of the present application.

[0023] Figure 5 A whole structure diagram of an LSTM-GA combined method is provided for an embodiment of the present application.

[0024] Figure 6 A flowchart of a scheduling optimization method of an electric vehicle virtual power plant is provided for an embodiment of the present application.

[0025] Figure 7 A structure schematic diagram of a computer device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0027] The above objects, features and advantages of the present application will become more apparent from the following detailed description considered in conjunction with the accompanying drawings and specific embodiments.

[0028] In one exemplary embodiment, as shown in Figure 1 A method for evaluating day-ahead response capability of an electric vehicle virtual power plant is provided, comprising the following steps 101-105.

[0029] Step 101, based on the historical operation data of each EV unit in the electric vehicle virtual power plant, the mean value calculation method is used to obtain the predicted value of the fixed characteristic parameters of each EV unit in the electric vehicle virtual power plant on the day to be predicted; the fixed characteristic parameters include: the allowable upper limit of battery state of charge, the allowable lower limit of battery state of charge and battery capacity.

[0030] Step 102, based on the historical operation data of each EV unit in the electric vehicle virtual power plant, the LSTM-GA combined method is used to obtain the predicted value of the non-fixed characteristic parameters of each EV unit in the electric vehicle virtual power plant on the day to be predicted; the non-fixed characteristic parameters include: the grid-connected state of charge, the minimum off-grid power, the grid-connected time and the off-grid time.

[0031] Step 103, according to the predicted value of each characteristic parameter of each EV unit in the electric vehicle virtual power plant on the day to be predicted, the clustering of each EV unit in the electric vehicle virtual power plant is carried out, and a plurality of clusters are obtained; each characteristic parameter includes each fixed characteristic parameter and each non-fixed characteristic parameter.

[0032] Step 104, the aggregation expression of each characteristic parameter in each cluster is calculated respectively.

[0033] Step 105, according to the aggregation expression of each characteristic parameter in each cluster, the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant are calculated.

[0034] The implementation of the above steps 101 to 105 can realize the evaluation of the day-ahead response capability of the electric vehicle virtual power plant.

[0035] In another exemplary embodiment of the present application, the step 105 of the present application is realized by the day-ahead response capability evaluation model constructed by the present application. First, the grid-connected response characteristics of the EV unit are modeled to obtain the day-ahead response capability of the EV unit under different response conditions, and the characteristic parameters affecting the grid-connected response capability of the EV unit are determined. Then, the characteristic parameters are used as clustering indexes to group the EVs, and the characteristic parameters are aggregated. Finally, the EV group is taken as the evaluation object to construct the day-ahead response capability evaluation model of the EVPP, which specifically includes the following steps 201-203.

[0036] Step 201: Calculation of the daytime response capability of the EV cell.

[0037] The response characteristics of independent EV cells when connected to the grid are as follows: Figure 2 As shown, the shaded area ABCFED is used to illustrate the maximum controllable charge and discharge range of an EV cell when connected to the grid: the EV cell at t a When connected to the grid, the state of charge is SOC. a (Point A) and immediately begin charging at the rated charging power. Once the maximum allowable state of charge (SOC) is reached, switch to idle mode. Figure 2 The middle boundary is ABC; while the boundary ADE represents the EV cell discharging at its rated discharge power immediately after being connected to the grid, when the state of charge drops to the minimum allowable charge level. SOC Discharge stops at a certain time, and the system switches to an idle state; to ensure the EV cells operate at t d The state of charge (SOC) when disconnected from the power grid is sufficient to meet the minimum energy requirements for the user's daily travel needs. d (This value is within the upper and lower limits of the allowable battery capacity). The EF segment indicates that the EV needs to be charged to meet the minimum battery capacity required for travel.

[0038] The entire process of a single EV cell being connected to the grid can be represented as the operating point (t, SOC) continuously moving over time within the maximum controllable charge and discharge range, and its actual trajectory depends on the cell's position within [t]. a Changes in grid connection status within the time period [t], such as Figure 1 As shown by the dashed line, after the EV enters the grid, it continuously changes its grid connection status, and its operating point movement trajectory changes accordingly, but it always remains within the maximum controllable area.

[0039] Taking electric vehicle j as an example, the change of battery state of charge during charging and discharging is shown in equation (1).

[0040]

[0041] In the formula: Let t be the state of charge of EV cell j at time t; and These represent the charging efficiency and discharging efficiency of EV cell j, respectively. Let be the charging and discharging power of EV cell j at time t-1 (with the power absorbed from the grid as the positive direction). In the charging state, it is equal to the rated charging power, and in the discharging state, it is equal to the negative value of the rated discharging power. δ represents the battery capacity of EV cell j; δ represents the length of a single evaluation period.

[0042] Without participating in any response (user daily charging), the EV does not have the ability to reverse discharge to the grid, i.e. there are only two grid-connected states of charging and idle. To better meet the travel demand, users will generally choose to charge immediately after the EV accesses the grid, and stop charging when the minimum off-grid power is reached to save charging costs, i.e. Figure 2 Trajectory A-G-F in FIG. 1. Therefore, without participating in the response, the grid-connected state of the jth EV The relationship between the state of charge at time t is shown in equation (2).

[0043]

[0044] In the equation: The values 0 and 1 respectively represent that the EV is in idle and charging state.

[0045] The day-ahead response capability evaluation of the EV unit refers to the estimation of the maximum response capability that the EV unit can provide at each evaluation time on the evaluation day without knowing the specific control strategy. When responding, the EV unit can change its grid-connected state in the non-response case to change the size of the grid-connected load. Defining the transition from idle state to discharge state as response case I, the transition from charging state to idle state as response case II, and the transition from idle state to charging state as case III, the first two response cases can achieve load reduction, and the third case can achieve load increase. It should be noted that when actually responding to the EV unit, considering the wear and tear of the EV battery, the charging and discharging power should not change too much, i.e. there is no response case of directly changing from charging state to discharging state.

[0046] Therefore, the day-ahead response capability of the EV unit can also be divided into the above three response cases, as shown in equations (3)-(5).

[0047]

[0048]

[0049]

[0050] In the equation: are the day-ahead response capabilities of the jth EV unit under the three response cases at time t, respectively; are the rated charging power and rated discharging power of the jth EV, respectively.

[0051] Step 202, EV group division.

[0052] In order to reduce the complexity of the final EVPP day-ahead response capability calculation, the embodiments of the present application first divide the EV monomers into groups, and aggregate EVs with similar response characteristics.

[0053] As can be seen from step 201, the day-ahead response capability of the EV monomer is limited by the maximum controllable area boundary of the grid connection, so the maximum controllable range of the grid connection can be used to distinguish EV monomers with different response characteristics.

[0054] As can be seen from formula (1) to formula (5), the parameters affecting the maximum controllable range of the grid connection of the EV monomer include: the off-grid time the upper limit of the battery state of charge the lower limit of the battery state of charge the grid-connected state of charge the minimum off-grid power and the battery rated capacity

[0055] The embodiments of the present application refer to the above-mentioned parameters as characteristic parameters of the EV monomer, and use the K-means method to realize the grouping of the EVs using the characteristic parameters as the clustering index, and use the elbow method to determine the optimal cluster number. The specific steps of K-means clustering are as follows:

[0056] (1) Collect the characteristic parameter data of all subordinate EV monomers in the EVPP to be evaluated on the day and normalize them as the basis for clustering.

[0057] (2) Take each EV monomer as a sample, take the size of the characteristic parameter as the sample coordinate, and randomly select K initial center points.

[0058] (3) Calculate the distance between the K center points and each sample using the Euclidean distance.

[0059] (4) Distribute each sample to the cluster containing the nearest center point.

[0060] (5) Calculate the coordinate mean of all samples in each cluster as the new center point of the cluster, and repeat steps (3) to (5) until the samples in each cluster no longer change.

[0061] The number of initial center points K is the final cluster number, and the elbow method can be used to determine the optimal cluster number.

[0062] The error sum of squares SSE is the core index of the elbow method, and its calculation method and theoretical basis are as follows:

[0063]

[0064] In the formula: k i is the number of sample points in the ith cluster; pij m represents the j-th sample point in the i-th cluster; i |p represents the center point of the i-th cluster; ij -m i | represents sample point p ij With center point m i The Euclidean distance between them; SSE is the clustering error of all samples, representing the quality of the clustering.

[0065] As the initial number of centroids K increases, the SSE of the corresponding clustering results can be calculated, and the results can be obtained. Figure 3 The relationship curve shown is shown in the figure.

[0066] When K is less than the optimal clustering number, increasing K significantly increases the aggregation degree of each cluster, resulting in a noticeable decrease in SSE. However, when K reaches the optimal clustering number, the return on aggregation degree from further increasing K decreases rapidly, and the decrease in SSE slows down sharply. This decrease then tends to level off as K continues to increase. In other words, the relationship between SSE and K resembles an elbow shape, with the elbow (e.g., ...) being the most important factor. Figure 3 The K value corresponding to the position of the solid point in the middle is the optimal number of clusters.

[0067] For the clustered EV group, the individual cells within it have similar battery characteristics and charging requirements. Therefore, the grid connection and disconnection time of the group and the characteristic parameters related to the state of charge can be represented by the geometric center (i.e., the mean) of the corresponding parameters of the individual cells, as shown in Equation (7). At the same time, the battery capacity and rated charge and discharge power of the EV group can also be expressed by the aggregated parameters of the corresponding parameters of the individual cells, as shown in Equation (8).

[0068]

[0069]

[0070] In the formula: This represents the characteristic parameters of the i-th EV population (i.e., the i-th cluster) after aggregation expression.

[0071] Step 203: Construct a day-ahead response capability assessment model for EVPP.

[0072] After clustering, the movement trajectory of each EV group's grid connection point under non-response conditions can be determined according to Equation (9), thus obtaining its grid connection status at each evaluation time. With state of charge

[0073]

[0074] in, Let represent the grid connection status of the i-th cluster at the evaluation time t on the day to be predicted. the state of charge of the i-th cluster at the evaluation time t of the day to be predicted, the aggregate expression of the on-grid time of the i-th cluster, the aggregate expression of the off-grid time of the i-th cluster, the aggregate expression of the minimum off-grid power of the i-th cluster, the aggregate expression of the allowable upper limit of the state of charge of the battery of the i-th cluster.

[0075] Then, the response capabilities of the EV population under the three response conditions are obtained according to formula (10), formula (11) and formula (12):

[0076]

[0077]

[0078]

[0079] wherein, and denote the response capabilities of the i-th cluster under response condition I, response condition II and response condition III at the evaluation time t of the day to be predicted, the aggregate expression of the rated charging power of the i-th cluster, the aggregate expression of the rated discharging power of the i-th cluster.

[0080] Finally, considering the response willingness degree of the EV users, the upper and lower boundaries of the day-ahead response capability of the EVPP can be obtained as shown in formula (13), and the actual response power of the EVPP will be constrained by the boundaries during the demand response scheduling.

[0081]

[0082] In the formula, the lower limit of the day-ahead response capability of the electric vehicle virtual power plant at the evaluation time t, the upper limit of the day-ahead response capability of the electric vehicle virtual power plant at the evaluation time t, the total charging power of the EV units in the electric vehicle virtual power plant at the evaluation time t without participation in response, and γ is the collective response willingness degree of the EV users, and N is the optimal clustering number.

[0083] Based on the clustering of the EVs, the upper and lower boundaries of the day-ahead response capability of the EVPP are finally obtained, and the power response range for the EVPP to participate in the demand response scheduling is determined. In order to ensure the rationality of the division of the EV population, i.e., always dividing the EV units with similar response characteristics into a class, an identification mechanism of the EV characteristic parameters needs to be established, so that the evaluation of the day-ahead response capability of the EVPP is more accurate.

[0084] In another exemplary embodiment of the present application, in steps 101 and 102, the characteristic parameters are divided into fixed characteristic parameters and non-fixed characteristic parameters, and the historical mean value is taken as the prediction value of the fixed characteristic parameters based on the historical operation data of the virtual power plant. For the non-fixed characteristic parameters, the LSTM-GA combined method is used to optimize the memory network hyperparameters, and then the final memory network is trained on the sample set to predict the parameters, which specifically includes steps 301-304.

[0085] Step 301, the size of the characteristic parameters of the EV monomer is mainly affected by the battery characteristics and user behavior, wherein the upper and lower limits of the battery state of charge and the battery capacity depend on the containment characteristics of the EV battery itself, and the value basically does not change, which can be considered as a fixed characteristic parameter (fixed parameter prediction), so the historical daily data of the previous year is selected to obtain the prediction value of the parameter value on the day to be evaluated.

[0086] The fixed characteristic parameter prediction of the jth EV monomer is shown in formula (14).

[0087]

[0088] In the formula: represents the fixed characteristic parameter on the Dth day in the past year.

[0089] Step 302, the non-fixed characteristic parameters are affected by the user's own travel and charging habits and have a certain volatility, which are non-fixed characteristic parameters: the size of the characteristic parameters of adjacent days will fluctuate, but the amplitude is not large. If the historical data sequence of these characteristic parameters is input into the prediction model and repeatedly trained, the long and short term time correlation of the time sequence can be effectively captured, and a more accurate prediction effect can be obtained.

[0090] The long short-term memory network (Long Short-Term Memory, LSTM) is a variant of a recurrent neural network (RNN), and the hidden layer thereof is composed of multiple state memory units adjacent to each other in time sequence, and the internal structure of a single state memory unit is as shown in Figure 4 , wherein x t is the input quantity at time step t, C t is the state information variable at time step t, and H tYtis the output variable at time step t. LSTM introduces a special storage unit and gating mechanism, which can capture and process long-term dependencies in sequence data, effectively solving the gradient vanishing and gradient exploding problems of traditional RNN when processing long sequences, thereby realizing accurate prediction of long time sequence data, so the application adopts the LSTM method to realize parameter identification.

[0091] In LSTM, the hidden layer is composed of a plurality of serially connected memory cells. Among them, the forgetting gate determines the forgetting part in the memory cell. x t and the output H t-1 of the previous time step are weighted and biased, and the information is transmitted through the sigmoid activation function σ(x), which outputs F t , F t closer to 1 indicates a higher degree of information retention of the previous memory cell, and closer to 0 indicates a higher degree of forgetting, and the functional relationship is represented as:

[0092] F t = σ(x t W xf + H t-1 W hf + b f ) (15)

[0093] In the formula: W xf and W hf are the weight values of the forgetting gate; b f is the bias value of the forgetting gate.

[0094] The input gate determines the reserved part in the memory cell and updates the cell information. x t and H t-1 are transmitted to the sigmoid activation function and the tanh activation function, producing I t and where is the initial state variable at time step t, I t is used to control the degree of superposition into the final state variable, and the larger the number indicates the higher the degree of superposition. Finally, I t and are multiplied, and the previous memory cell information is selectively recorded into the current time state information variable according to F t , which is represented as:

[0095] I t = σ(x t W xi + H t-1 W hi + b i ) (16)

[0096]

[0097]

[0098] In the formula: W xi and W hi b represents the weight values ​​of the input gate. i This is the bias value of the input gate.

[0099] The output gate determines the output of the current state memory unit, expressed as:

[0100] O t =σ(x t W xo +H t-1 W ho +b o (19)

[0101] H t =O t ×tanh(C t (20)

[0102] In the formula: W xo and W ho b represents the weight value of the output gate; o This is the bias value for the output gate.

[0103] Prediction of non-fixed characteristic parameters, specific scheme: Based on the non-fixed characteristic parameters... Taking the example of day-ahead parameter identification, firstly, historical daily data of the state of charge of the j-th EV unit in the past year are collected as known data. Abnormal data is removed and the time series is completed using interpolation. The processed data is then normalized, and then a certain time window size t is used. w Rolling to generate training samples: t w The data from day t is used as the input sequence of a training sample. w +1 day Data prediction can be represented as:

[0104]

[0105]

[0106] In the formula: X (n) Y (n) These are the input sequence and the quantity to be predicted in the nth training sample, respectively. This represents the grid-connected charge status of the j-th EV on day n.

[0107] If a year has 365 days, then the total amount can be 365-t. wGroup time series samples, divide the total samples into training set and test set in the ratio of 7:3, train the network using the training set, and use mean square error (MSE) on the test set to evaluate the performance of the prediction model, the calculation method of MSE is shown in formula (24):

[0108]

[0109] In the formula: N t is the number of test set samples; respectively, the actual value and the predicted value of the kth test set sample to be predicted day ; the smaller the MSE value, the better the prediction ability of the LSTM model.

[0110] The GA method realizes the optimization of the hyperparameters of the LSTM network. The size of the time window, the number of state memory units contained in the hidden layer, and other hyperparameters will affect the prediction accuracy of the LSTM network. Manual parameter adjustment each time will waste a lot of time and the tuning effect is poor, therefore, the genetic algorithm (Genetic Algorithm, GA) is used to optimize the hyperparameters of the LSTM network to obtain the hyperparameter combination with the highest prediction accuracy. The overall structure of the LSTM-GA combination method is shown in Figure 5 , and the optimization process is as follows:

[0111] (1) Initialize the population: take the time window size and the number of state memory units of the LSTM network as the optimization variables (genetic information), set the upper and lower thresholds respectively, and randomly generate m groups of hyperparameter combinations as the initial population of the genetic algorithm;

[0112] (2) Genetic information coding: encode the genetic information, and the binary coding is used in the present application;

[0113] (3) Calculate the fitness: decode the binary coding of each population individual into the hyperparameter combination represented, respectively, into the LSTM network and train on the training set, and calculate 1 / MSE on the test set as the individual fitness;

[0114] (4) Roulette wheel selection: according to the size of the individual fitness, each individual is given a selection probability, and a individual is randomly selected as the parent according to the selection probability, then the individual is excluded and the roulette wheel selection is used to select the parent, until m / 2 parents are selected, then stop;

[0115] (5) Crossover and mutation: pair the parents two by two, exchange the genetic information (binary coding) of the two according to certain rules to generate m / 2 new individuals, and perform random mutation operation (0-1 replacement) on the genetic information (binary coding) of the new individuals to increase the diversity of the genetic information;

[0116] (6) combine the mutated new individual with the parent to obtain a new generation population;

[0117] (7) repeat steps (2)-(6) until the optimal individual fitness no longer changes or the maximum number of iterations is reached, then stop iteration, input the hyperparameters represented by the current optimal individual into the LSTM network, and train the optimal parameter model on the entire sample set.

[0118] After training, the day-ahead t w The day-ahead t forecast value can be output by inputting the optimal parameter LSTM network with the day-ahead t

[0119] Similarly, the identification of other non-fixed characteristic parameters can also be achieved by the combined method.

[0120] In combination with the above day-ahead response capability evaluation model, according to the identification results of the characteristic parameters in steps 102 and 103, the EVs are grouped using step 103, and the upper and lower limits of the day-ahead response capability of the EVPP are obtained from equation (13).

[0121] In an exemplary embodiment, as shown in Figure 6 , a scheduling optimization method for an electric vehicle virtual power plant is provided, including the following steps 601-603.

[0122] Step 601, an optimization model is constructed with the goal of optimal tracking effect of actual response power on scheduling demand signal and minimum amount of abandoned wind and light.

[0123] Step 602, based on the upper and lower limits of the day-ahead response capability of the electric vehicle virtual power plant, the constraint conditions of the optimization model are determined; the upper and lower limits of the day-ahead response capability of the electric vehicle virtual power plant are obtained using the day-ahead response capability evaluation method of the electric vehicle virtual power plant described above.

[0124] Step 603, based on the constraint conditions, the optimization model is solved to obtain an optimal scheduling scheme.

[0125] In another exemplary embodiment, the EVPP participates in demand response by responding to the scheduling demand signal (including the time period that needs to be responded to and the power size that needs to be responded to) issued by the superior dispatching center. The present embodiment establishes an optimization model with the goal of optimal improvement effect of the EVPP on the system load curve (tracking effect of actual response power on scheduling demand signal) and minimum amount of abandoned wind and light:

[0126] minF=min(ω1f1+ω2f2) (24)

[0127]

[0128]

[0129]

[0130]

[0131] wherein F is a minimum objective function of the optimization model; f1 and f2 are two sub-objective functions of the optimization model, and f1 and f2 are both minimum, so formula (24) is a form of addition of the two; ω1 and ω2 are weight coefficients for de-dimensioning; T DR is a time period in which the upper dispatching center needs the electric vehicle virtual power plant to respond; is an actual operation power of the electric vehicle virtual power plant at the dispatching time t'; is a response power size of the electric vehicle virtual power plant to the demand of the upper dispatching center at the dispatching time t'; respectively are abandoned photovoltaic power generation and abandoned wind power generation; respectively are a predicted output of photovoltaic power generation and an actual output for balancing load at the dispatching time t', and a difference between the two is abandoned photovoltaic power at the time; respectively are a predicted output of wind power generation and an actual output for balancing load at the dispatching time t', and a difference between the two is abandoned wind power at the time.

[0132] The optimization variable of the optimization model is an operation power of the EVPP at each dispatching time and an actual output of the distributed new energy at each dispatching time, and the remaining variables have been determined by the upper dispatching center through optimization.

[0133] The constraint condition of the model can be expressed as follows:

[0134]

[0135]

[0136]

[0137]

[0138] wherein, is a channel transmission power of the upper power grid at the dispatching time t', is a basic load of the region where the electric vehicle virtual power plant is located at the dispatching time t', is an actual operation power of the electric vehicle virtual power plant at the dispatching time t', is a lower limit of the day-ahead response capability of the electric vehicle virtual power plant at the dispatching time t', an upper limit of the day-ahead response capability of the electric vehicle virtual power plant at the dispatch time t', an actual operation power of the electric vehicle virtual power plant at the dispatch time t', a predicted output of photovoltaic power generation and an actual output for balancing load at the dispatch time t' respectively, a predicted output of wind power generation and an actual output for balancing load at the dispatch time t' respectively. Further, the above formula (29) is a system power balance constraint, are determined by the superior dispatch center; the above formula (30) is an EVPP output constraint, indicating that the actual output of the EVPP should always be within the upper and lower boundaries of the day-ahead response capability; and the above formulas (31) and (32) are output constraints of distributed photovoltaic power generation and distributed wind power generation respectively.

[0139] In actual dispatch optimization, the sizes of ω1 and ω2 can be flexibly set according to the priority of the system dispatch demand: when more emphasis is placed on improving the system load curve, i.e., the EVPP needs to achieve better tracking of the dispatch demand signal, the size of ω1 can be appropriately increased; when more emphasis is placed on improving the utilization rate of new energy, i.e., reducing the amount of abandoned wind and light, the size of ω2 can be appropriately increased.

[0140] The period T DR For part of the period in the day to be predicted in the embodiment of the day-ahead response capability evaluation method of the electric vehicle virtual power plant, in the specific implementation process, the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant in the above step 602 can be obtained by the following steps:

[0141] First, the period T DR to be predicted, then the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant at each evaluation time in the day to be predicted are obtained by using the day-ahead response capability evaluation method of the electric vehicle virtual power plant in the above embodiment, and then the period T DR the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant at each dispatch time t' in the period are selected. In the application embodiment, the dispatch times and the evaluation times in the same period can be one-to-one corresponding, or the dispatch times can correspond to part of the evaluation times, for example, the time interval of the evaluation times is 1s, and the time interval of the dispatch times can be an integer multiple of 1s.

[0142] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and an internal structure diagram thereof can be as follows:Figure 7 The computer device shown in the figure includes a processor, a memory, an input / output interface (I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement the day-ahead response capability evaluation method of the electric vehicle virtual power plant or the scheduling optimization method of the electric vehicle virtual power plant.

[0143] Those skilled in the art can understand that, Figure 7 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement. In one exemplary embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in each of the method embodiments described above.

[0144] According to the specific embodiments provided by the present application, the present application has the following technical effects.

[0145] The present application adopts the K-means clustering method to group all EVs in the virtual power plant by taking characteristic parameters as clustering indicators, and determines the optimal clustering number by the elbow method, realizes the aggregation expression of EVs with similar response characteristics, and establishes an evaluation model for the day-ahead response capability of the EVPP by taking the EV group as the evaluation object, which reduces the calculation complexity of the evaluation model, and the EV group can also be taken as the optimization object to participate in the scheduling optimization of the EVPP to improve the optimization speed.

[0146] 2、The present application is based on the idea of data-driven, and adopts the LSTM-GA combined method to capture the time dependence between the historical operation data of the virtual power plant, realizes the accurate identification of the non-fixed characteristic parameters of the EV individual, and ensures the optimal prediction network hyperparameters, which more accurately reflects the differences between the EV individuals in the actual scene.

[0147] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties.

[0148] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiments of each method. Any reference to memory, database or other medium used in the embodiments provided by the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (Read-Only Memory, ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), etc.

[0149] The database involved in the embodiments provided by the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided by the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0150] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combination of the technical features does not exist contradictory, it should be considered as the scope of the present application.

[0151] The principles and implementation manners of the present application are described herein by using specific examples, and the above examples are only used to help understand the method of the present application and its core idea; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manners and application ranges will have changes. In conclusion, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A method for evaluating day-ahead response capability of an electric vehicle virtual power plant, characterized in that, The method comprises the following steps: Based on the historical operation data of each EV unit in the electric vehicle virtual power plant, the mean value calculation method is used to obtain the predicted value of the fixed characteristic parameter of each EV unit in the electric vehicle virtual power plant on the day to be predicted; the fixed characteristic parameter includes: the allowable upper limit of battery state of charge, the allowable lower limit of battery state of charge and battery capacity; Based on the historical operation data of each EV unit in the electric vehicle virtual power plant, the LSTM-GA combined method is used to obtain the predicted value of the non-fixed characteristic parameter of each EV unit in the electric vehicle virtual power plant on the day to be predicted; the non-fixed characteristic parameter includes: the grid-connected state of charge, the minimum off-grid power, the grid-connected time and the off-grid time; According to the predicted value of each characteristic parameter of each EV unit in the electric vehicle virtual power plant on the day to be predicted, each EV unit in the electric vehicle virtual power plant is clustered to obtain a plurality of clusters; each characteristic parameter includes each fixed characteristic parameter and each non-fixed characteristic parameter; The aggregation expression of each characteristic parameter in each cluster is calculated respectively; According to the aggregation expression of each characteristic parameter in each cluster, the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant are calculated, which specifically includes: According to the aggregation expression of each characteristic parameter in each cluster, the grid-connected state of each cluster in the electric vehicle virtual power plant at each evaluation time on the day to be predicted is calculated by using the following formula: ； wherein, is the grid-connected state of the i-th cluster at the evaluation time t of the day to be predicted, is the state of charge of the i-th cluster at the evaluation time t of the day to be predicted, is the aggregate expression of the time of entry into the grid of the i-th cluster, is the aggregate expression of the time of exit from the grid of the i-th cluster, is the aggregate expression of the minimum amount of electricity to exit from the grid of the i-th cluster, is the aggregate expression of the allowed upper limit of the state of charge of the battery of the i-th cluster; According to the aggregation expression of each characteristic parameter in each cluster and the grid-connected state of each cluster at each evaluation time on the day to be predicted, the response capability of each cluster in different response situations at each evaluation time on the day to be predicted is calculated by using the following formula: ； ； ； wherein, , and respectively represent the response capability of the response case I, the response case II and the response case III of the ith cluster at the evaluation time t of the day to be predicted, is an aggregate expression of the rated charging power of the ith cluster, is an aggregate expression of the rated discharging power of the ith cluster; the response case I is the case of transition from the idle state to the discharging state, the response case II is the case of transition from the charging state to the idle state, and the response case III is the case of transition from the idle state to the charging state; According to the aggregation expression of each characteristic parameter in each cluster and the response capability of each cluster in different response situations at each evaluation time on the day to be predicted, the upper limit and the lower limit of the day-ahead response capability of the electric vehicle virtual power plant are calculated by using the following formula: ； wherein, is the lower limit of the day-ahead response capability of the EV virtual power plant at time t, is the upper limit of the day-ahead response capability of the EV virtual power plant at time t, is the total charging power of the EVs in the EV virtual power plant at time t when no response is involved, is the collective response willingness of the EV users, is the optimal number of clusters. 2.The day-ahead response capability evaluation method of an electric vehicle virtual power plant according to claim 1, characterized in that, Based on the historical operation data of each EV unit in the electric vehicle virtual power plant, the mean value calculation method is used to obtain the predicted value of the fixed characteristic parameter of each EV unit in the electric vehicle virtual power plant on the day to be predicted, which specifically includes: Based on the historical operation data of each EV unit in the electric vehicle virtual power plant, the predicted value of the fixed characteristic parameter of each EV unit in the electric vehicle virtual power plant on the day to be predicted is obtained by using the following formula: ； in, , and These represent the upper limit of the allowable state of charge (SOC) of EV cell j on the predicted date, the lower limit of the allowable SOC, and the predicted value of the battery capacity. , and For each of the EV monomers j in the past year The upper limit of the permissible state of charge (SOC) of the battery, the lower limit of the permissible SOC of the battery, and the actual value of the battery capacity. 3.The day-ahead response capability evaluation method of the electric vehicle virtual power plant according to claim 1, wherein, Based on the historical operation data of each EV unit in the electric vehicle virtual power plant, the LSTM-GA combined method is used to obtain the predicted value of the non-fixed characteristic parameter of each EV unit in the electric vehicle virtual power plant on the day to be predicted, which specifically includes: Based on the actual value of the mth non-fixed characteristic parameter of each EV unit in the electric vehicle virtual power plant in the past year, a training sample is generated by using a time window roll; m=1, 2, …, M, M is the number of non-fixed characteristic parameters; Each training sample is divided into a training set and a test set; Based on the training set and the test set, the mean square error is used as a loss function, the GA algorithm is used to train the LSTM network model, and a trained LSTM network model is obtained. The EV monomer j will be as follows before the forecast date. The actual value of the m-th non-fixed characteristic parameter of a given day is input into the trained LSTM network model to obtain the predicted value of the m-th non-fixed characteristic parameter of EV unit j on the day to be predicted, where j = 1, 2, ..., J, and J is the number of EV units in the virtual electric vehicle power plant. This represents the size of the time window. 4.The method of claim 1, wherein, According to the predicted values of the characteristic parameters of each EV unit in the electric vehicle virtual power plant on the day to be predicted, the electric vehicle virtual power plant is clustered, and a plurality of clusters are obtained, specifically including: Initializing the number K of cluster centers; According to the predicted values of the characteristic parameters of each EV unit in the electric vehicle virtual power plant on the day to be predicted, the electric vehicle virtual power plant is clustered using the K-means clustering algorithm, and K clusters are obtained; Calculate the sum of squared errors of the K clusters; Based on the elbow method, determine whether the absolute value of the difference between the sum of squared errors of the K clusters obtained in the current iteration and the sum of squared errors of the K-1 clusters obtained in the previous iteration is greater than a preset threshold, and obtain a judgment result; If the judgment result is yes, increase the value of K by 1, and return to the step of "according to the predicted values of the characteristic parameters of each EV unit in the electric vehicle virtual power plant on the day to be predicted, using the K-means clustering algorithm to cluster each EV unit in the electric vehicle virtual power plant, and obtaining K clusters"; If the judgment result is no, the value of K in the current iteration is taken as the optimal clustering number, and the K clusters obtained in the current iteration are output.

5. The method for evaluating day-ahead response capability of an electric vehicle virtual power plant according to claim 4, wherein, The calculation formula of the sum of squared errors is: ； wherein, is the sum of squared errors, is the parameter vector of the EV individual j in the i-th cluster, is the parameter vector of the cluster center of the i-th cluster, is the number of EV individuals in the i-th cluster; the parameter vector of the EV individual j consists of the predicted values of the characteristic parameters of the EV individual j on the day to be predicted, and the parameter vector of the cluster center consists of the average values of the predicted values of the characteristic parameters of each EV individual in the i-th cluster on the day to be predicted.

6. A scheduling optimization method of an electric vehicle virtual power plant, characterized by, including: An optimization model is constructed with the optimal tracking effect of the actual response power on the scheduling demand signal and the minimum amount of abandoned wind and light as the target; Based on the upper and lower limits of the day-ahead response capacity of the electric vehicle virtual power plant, the constraint conditions of the optimization model are determined; The upper and lower limits of the day-ahead response capacity of the electric vehicle virtual power plant are obtained by the day-ahead response capacity evaluation method of the electric vehicle virtual power plant in any one of claims 1-5; Based on the constraint conditions, the optimization model is solved to obtain an optimal scheduling scheme. 7.The scheduling optimization method of the electric vehicle virtual power plant according to claim 6, characterized in that, The optimization model is: ； ； ； ； ； wherein, is a minimum objective function of the optimization model; , are two sub-objective functions of the optimization model, respectively; , is a weight coefficient for de-dimensioning; is a time period in which the superior dispatching center needs the electric vehicle virtual power plant to respond; is a dispatching time is an actual operating power of the electric vehicle virtual power plant; is a dispatching time is a response power size of the electric vehicle virtual power plant to the demand of the superior dispatching center; , are abandoned photovoltaic power generation and abandoned wind power generation, respectively; , are a predicted output of photovoltaic power generation and an actual output for balancing load at a dispatching time , are a predicted output of wind power generation and an actual output for balancing load at a dispatching time , . 8.The scheduling optimization method of the electric vehicle virtual power plant according to claim 6, wherein, The constraint condition is: ； ； ； ； wherein, the dispatch time the transmission power of the channel of the upper-level power grid, the dispatch time the basic load of the region where the electric vehicle virtual power plant is located, the dispatch time the actual operating power of the electric vehicle virtual power plant, the dispatch time the lower limit of the day-ahead response capability of the electric vehicle virtual power plant, the dispatch time the upper limit of the day-ahead response capability of the electric vehicle virtual power plant, , the dispatch time the predicted output of photovoltaic power generation and the actual output for balancing the load, , the dispatch time the predicted output of wind power generation and the actual output for balancing the load.

9. A computer device comprising: A memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that the processor executes the computer program to implement the day-ahead response capacity evaluation method of the electric vehicle virtual power plant in any one of claims 1-5, or the scheduling optimization method of the electric vehicle virtual power plant in any one of claims 6-8.

Images