Electric vehicle reserve capacity prediction method and device, equipment and storage medium

Abstract

The invention discloses an electric vehicle reserve capacity prediction method, device and equipment and a storage medium, and the method comprises the steps: collecting historical charging data and related behavior data of an electric vehicle, carrying out the cleaning and standardization processing of the historical charging data and the behavior data, and obtaining the processed target data; predicting future behavior information of the electric vehicle according to the target data by using a long short-term memory (LSTM) network, and analyzing intention information of users of the electric vehicle participating in centralized regulation and control according to the target data by using a fuzzy neural network (FNN); according to future behavior information and willingness information, an LSTM-FNN model is trained and optimized, the reserve capacity of the electric vehicle is obtained through calculation, the reserve capacity is analyzed, a target function of corresponding power linear programming is obtained, the uncertainty of user participation willingness can be considered, the precision and stability of reserve capacity prediction are improved, and the prediction efficiency is improved. And the operation efficiency and reliability of the power system can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle auxiliary services, and in particular to a method, device, equipment and storage medium for predicting the spare capacity of an electric vehicle. Background Art

[0002] With their fast and accurate dynamic response capabilities and considerable cluster capacity, electric vehicles have become a new idea for solving the backup problem of power systems; however, the battery capacity of a single electric vehicle is small and its driving is highly random. Existing studies usually use aggregation models to predict its backup capacity, but do not consider the uncertainty of user participation willingness. Summary of the invention

[0003] The main purpose of the present invention is to provide a method, device, equipment and storage medium for predicting the spare capacity of an electric vehicle, aiming to solve the technical problem that the prior art uses an aggregation model to predict the spare capacity, but does not take into account the uncertainty of user participation willingness, has prediction errors, and leads to inaccurate capacity prediction.

[0004] In a first aspect, the present invention provides a method for predicting the spare capacity of an electric vehicle, the method comprising the following steps:

[0005] Collecting historical charging data and related behavior data of electric vehicles, cleaning and normalizing the historical charging data and the behavior data, and obtaining processed target data;

[0006] Using a long short-term memory network LSTM to predict future behavior information of the electric vehicle according to the target data, and using a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data;

[0007] The LSTM-FNN model is trained and optimized according to the future behavior information and the intention information, the spare capacity of the electric vehicle is calculated, the spare capacity is analyzed, and the objective function of the corresponding power linear programming is obtained.

[0008] Optionally, the collecting of historical charging data and related behavior data of the electric vehicle, cleaning and normalizing the historical charging data and the behavior data, and obtaining processed target data include:

[0009] Collecting the arrival time, departure time, charging state, expected power and related behavior data of the electric vehicle in a historical time period, and using the arrival time, departure time, charging state and expected power as historical charging data;

[0010] Performing data cleaning on the historical charging data and the behavior data to remove missing values ​​and abnormal values ​​in the historical charging data and the behavior data;

[0011] The cleaned data is normalized to obtain the processed target data.

[0012] Optionally, the using of a long short-term memory network LSTM to predict future behavior information of the electric vehicle according to the target data, and using a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data, includes:

[0013] Determine a training set and a test set corresponding to the target data according to a preset ratio;

[0014] Using a long short-term memory network (LSTM) to predict and obtain future behavior information of the electric vehicle based on the training set and the test set;

[0015] The grid connection time, SOC and electricity price are obtained from the target data, the grid connection time, SOC and electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

[0016] Optionally, the using a long short-term memory network (LSTM) to predict and obtain the future behavior information of the electric vehicle according to the training set and the test set includes:

[0017] The training set and the test set are input into the long short-term memory network LSTM to obtain the network output result

[0018] Calculate the mean square error between the network output result and the target value in the training set, and use the mean square error as the training loss function;

[0019] Obtain an observed value from the network output result, and take the square root of the second-order sample moment of the difference between the predicted value of the test set and the observed value as the root mean square error;

[0020] Pruning the LSTM model according to the training loss function and the root mean square error to obtain a pruned target LSTM model;

[0021] The target data is predicted according to the target LSTM model to obtain future behavior information of the electric vehicle.

[0022] Optionally, the grid connection time, SOC and electricity price are obtained from the target data, the grid connection time, SOC and electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output, including:

[0023] The grid connection time is obtained from the target data by the following formula:

[0024] t=t step -t arr +t dep

[0025] Among them, t is the grid connection time, t step is the time step, representing a specific time, t arr is the arrival time of the electric vehicle, t dep Leave time for electric cars;

[0026] The state of charge SOC and the electricity price are obtained from the target data, the grid connection time, the SOC and the electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

[0027] Optionally, obtaining the state of charge SOC and the electricity price from the target data, inputting the grid connection time, the SOC and the electricity price into a fuzzy neural network FNN, and outputting the willingness information of the user of the electric vehicle to participate in centralized regulation, includes:

[0028] Using charge prediction data from the target data as the state of charge SOC;

[0029] The income after the electric vehicle participates in the dispatch, the self-charging income, the reserve income, the emission income and the charging cost corresponding to the electricity price are obtained from the target data, and the profit loss of the electric vehicle user participating in the centralized regulation is calculated by the following formula based on the income after the dispatch, the self-charging income, the reserve income, the emission income and the charging cost:

[0030]

[0031] Among them, a is the profit loss, M L is the income after participating in the scheduling, M B is the self-charging income, M reserve For reserve income,

[0032] Determine the willingness information of the user of the electric vehicle to participate in the centralized regulation according to the profit loss, M c For charging cost.

[0033] Optionally, the training and optimizing of the LSTM-FNN model according to the future behavior information and the willingness information, calculating and obtaining the spare capacity of the electric vehicle, analyzing the spare capacity, and obtaining the objective function of the corresponding power linear programming include:

[0034] The LSTM-FNN model is trained and optimized according to the future behavior information and the willingness information, and the spare capacity of the electric vehicle is obtained by the following calculation:

[0035]

[0036]

[0037] Among them, P cu (k) is the upper limit reserve capacity, γ(k) is the adjustment coefficient, P min is the minimum discharge power, P(k) is the operating power, E(k) is the battery capacity of the electric vehicle, and E min (k+1) is the minimum power requirement for the next time step k+1, P cd (k) is the lower limit reserve capacity, P max is the maximum charging power, E max is the maximum battery capacity, Δt is the length of the time step;

[0038] Correspondingly, the constraints are:

[0039] P cu (t)≤-P min +P(t)

[0040] P cd (t)≤P max -P(t)

[0041] Among them, P cu (t) is the upper limit reserve capacity, P cd (t) is the lower limit reserve capacity, P(t) is the operating power, P min is the minimum discharge power, P max is the maximum charging power;

[0042]

[0043] Among them, E min is the battery minimum charging power constraint, E base is the minimum power of electric vehicles, E exp is the expected power, P max is the maximum charging power, t dep is the departure time, t is the current time, t base is the base time;

[0044]

[0045] P min ≤P(t)≤P max

[0046] i c(t)+i d (t)≤1

[0047] Where, P(t) is the operating power, P c (t) is the charging power, P d (t) is the discharge power, i c (t) is the 0-1 charge-discharge state variable, i d (t) is the 0-1 charge-discharge state variable, i c (t) = 0 means the electric vehicle is not charged, i c (t) = 1 means the electric vehicle has been charged. For charging efficiency, is the discharge efficiency, P max is the maximum charging power, P min is the minimum discharge power;

[0048] The reserve capacity is analyzed and the objective function of the corresponding power linear programming is obtained by the following formula:

[0049]

[0050] M=M r.opt +M d.opt -M c.opt

[0051] M r.opt =P cu ×π cu

[0052] M d.opt =P cd ×π cd

[0053] M c.opt =P d ×π d -P c ×π c

[0054]

[0055] Among them, M agg For aggregators to control the overall revenue of electric vehicles, M j is the total income of electric vehicles j participating in the aggregate regulation, M is the objective function, M r.opt is the reserve income of electric vehicles after participating in the aggregator dispatch, M d.opt is the emission income of electric vehicles after participating in the aggregator dispatch, M c.opt is the charging cost of electric vehicles after participating in the aggregator dispatch, π cu , π cd , π c , πd are the prices of upper reserve, lower reserve, charging power and discharging power respectively, N is the total number of electric vehicles in the cluster, The electric vehicle does not accept the charging and discharging power of centralized scheduling, P cu , P cd , P c , P d It is an intermediate variable related to power P.

[0056] In a second aspect, in order to achieve the above-mentioned purpose, the present invention further proposes a device for predicting the spare capacity of an electric vehicle, the device comprising:

[0057] A data processing module, used to collect historical charging data and related behavior data of the electric vehicle, clean and normalize the historical charging data and the behavior data, and obtain processed target data;

[0058] An information analysis module, used to predict the future behavior information of the electric vehicle according to the target data using a long short-term memory network LSTM, and to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data using a fuzzy neural network FNN;

[0059] The prediction and planning module is used to train and optimize the LSTM-FNN model according to the future behavior information and the willingness information, calculate the spare capacity of the electric vehicle, analyze the spare capacity, and obtain the objective function of the corresponding power linear programming.

[0060] In a third aspect, in order to achieve the above-mentioned purpose, the present invention also proposes an electric vehicle reserve capacity prediction device, which includes: a memory, a processor, and an electric vehicle reserve capacity prediction program stored in the memory and executable on the processor, and the electric vehicle reserve capacity prediction program is configured to implement the steps of the electric vehicle reserve capacity prediction method as described above.

[0061] In a fourth aspect, in order to achieve the above-mentioned purpose, the present invention further proposes a storage medium, on which an electric vehicle reserve capacity prediction program is stored, and when the electric vehicle reserve capacity prediction program is executed by a processor, the steps of the electric vehicle reserve capacity prediction method as described above are implemented.

[0062] The electric vehicle reserve capacity prediction method proposed in the present invention collects historical charging data and related behavior data of the electric vehicle, cleans and normalizes the historical charging data and the behavior data, and obtains processed target data; uses a long short-term memory network LSTM to predict future behavior information of the electric vehicle according to the target data, and uses a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data; trains and optimizes the LSTM-FNN model according to the future behavior information and the willingness information, calculates and obtains the reserve capacity of the electric vehicle, analyzes the reserve capacity, and obtains the objective function of the corresponding power linear programming, which can take into account the uncertainty of the user's willingness to participate, improves the accuracy and stability of the reserve capacity prediction, and helps to improve the efficiency and reliability of the power system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention;

[0064] Figure 2 It is a flow chart of a first embodiment of a method for predicting spare capacity of an electric vehicle according to the present invention;

[0065] Figure 3 A schematic diagram of a flow chart of a second embodiment of a method for predicting spare capacity of an electric vehicle according to the present invention;

[0066] Figure 4 A schematic diagram of a flow chart of a third embodiment of a method for predicting spare capacity of an electric vehicle according to the present invention;

[0067] Figure 5 A schematic diagram of a flow chart of a fourth embodiment of a method for predicting spare capacity of an electric vehicle according to the present invention;

[0068] Figure 6 This is a functional module diagram of the first embodiment of the electric vehicle spare capacity prediction device of the present invention.

[0069] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0070] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0071] The solution of the embodiment of the present invention is mainly: by collecting historical charging data and related behavior data of electric vehicles, cleaning and normalizing the historical charging data and the behavior data, and obtaining processed target data; using a long short-term memory network LSTM to predict the future behavior information of the electric vehicle according to the target data, and using a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data; training and optimizing the LSTM-FNN model according to the future behavior information and the willingness information, calculating and obtaining the spare capacity of the electric vehicle, analyzing the spare capacity, and obtaining the objective function of the corresponding power linear programming, which can take into account the uncertainty of the user's willingness to participate, improve the accuracy and stability of the spare capacity prediction, and help improve the efficiency and reliability of the power system operation, and solve the technical problem that the aggregation model is used to predict the spare capacity in the prior art, but the uncertainty of the user's willingness to participate is not considered, there is a prediction error, and the capacity prediction is inaccurate.

[0072] Reference Figure 1 , Figure 1 The figure is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention.

[0073] like Figure 1 As shown, the device may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a Wi-Fi interface). The memory 1005 may be a high-speed RAM memory, or it may be a stable memory (Non-Volatile Memory), such as a disk memory. The memory 1005 may optionally be a storage device independent of the aforementioned processor 1001.

[0074] Those skilled in the art will understand that Figure 1 The device structure shown in the figure does not constitute a limitation of the device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0075] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating device, a network communication module, a user interface module, and an electric vehicle reserve capacity prediction program.

[0076] The device of the present invention calls the electric vehicle reserve capacity prediction program stored in the memory 1005 through the processor 1001, and performs the following operations:

[0077] Collecting historical charging data and related behavior data of electric vehicles, cleaning and normalizing the historical charging data and the behavior data, and obtaining processed target data;

[0078] Using a long short-term memory network LSTM to predict future behavior information of the electric vehicle according to the target data, and using a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data;

[0079] The LSTM-FNN model is trained and optimized according to the future behavior information and the intention information, the spare capacity of the electric vehicle is calculated, the spare capacity is analyzed, and the objective function of the corresponding power linear programming is obtained.

[0080] The device of the present invention calls the electric vehicle reserve capacity prediction program stored in the memory 1005 through the processor 1001, and also performs the following operations:

[0081] Collecting the arrival time, departure time, charging state, expected power and related behavior data of the electric vehicle in a historical time period, and using the arrival time, departure time, charging state and expected power as historical charging data;

[0082] Performing data cleaning on the historical charging data and the behavior data to remove missing values ​​and abnormal values ​​in the historical charging data and the behavior data;

[0083] The cleaned data is normalized to obtain the processed target data.

[0084] The device of the present invention calls the electric vehicle reserve capacity prediction program stored in the memory 1005 through the processor 1001, and also performs the following operations:

[0085] Determine a training set and a test set corresponding to the target data according to a preset ratio;

[0086] Using a long short-term memory network (LSTM) to predict and obtain future behavior information of the electric vehicle based on the training set and the test set;

[0087] The grid connection time, SOC and electricity price are obtained from the target data, the grid connection time, SOC and electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

[0088] The device of the present invention calls the electric vehicle reserve capacity prediction program stored in the memory 1005 through the processor 1001, and also performs the following operations:

[0089] The training set and the test set are input into the long short-term memory network LSTM to obtain the network output result

[0090] Calculate the mean square error between the network output result and the target value in the training set, and use the mean square error as the training loss function;

[0091] Obtain an observed value from the network output result, and take the square root of the second-order sample moment of the difference between the predicted value of the test set and the observed value as the root mean square error;

[0092] Pruning the LSTM model according to the training loss function and the root mean square error to obtain a pruned target LSTM model;

[0093] The target data is predicted according to the target LSTM model to obtain future behavior information of the electric vehicle.

[0094] The device of the present invention calls the electric vehicle reserve capacity prediction program stored in the memory 1005 through the processor 1001, and also performs the following operations:

[0095] The grid connection time is obtained from the target data by the following formula:

[0096] t=t step -t arr +t dep

[0097] Among them, t is the grid connection time, t step is the time step, representing a specific time, t arr is the arrival time of the electric vehicle, t dep Leave time for electric cars;

[0098] The state of charge SOC and the electricity price are obtained from the target data, the grid connection time, the SOC and the electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

[0099] The device of the present invention calls the electric vehicle reserve capacity prediction program stored in the memory 1005 through the processor 1001, and also performs the following operations:

[0100] Using charge prediction data from the target data as the state of charge SOC;

[0101] The income after the electric vehicle participates in the dispatch, the self-charging income, the reserve income, the emission income and the charging cost corresponding to the electricity price are obtained from the target data, and the profit loss of the electric vehicle user participating in the centralized regulation is calculated by the following formula based on the income after the dispatch, the self-charging income, the reserve income, the emission income and the charging cost:

[0102]

[0103] Among them, a is the profit loss, M L is the income after participating in the scheduling, M B is the self-charging income, M reserve For reserve income,

[0104] Determine the willingness information of the user of the electric vehicle to participate in the centralized regulation according to the profit loss, M c For charging cost.

[0105] The device of the present invention calls the electric vehicle reserve capacity prediction program stored in the memory 1005 through the processor 1001, and also performs the following operations:

[0106] The LSTM-FNN model is trained and optimized according to the future behavior information and the willingness information, and the spare capacity of the electric vehicle is obtained by the following calculation:

[0107]

[0108] Among them, P cu (k) is the upper limit reserve capacity, γ(k) is the adjustment coefficient, P min is the minimum discharge power, P(k) is the operating power, E(k) is the battery capacity of the electric vehicle, and E min (k+1) is the minimum power requirement for the next time step k+1, P cd (k) is the lower limit reserve capacity, P max is the maximum charging power, E max is the maximum battery capacity, Δt is the length of the time step;

[0109] Correspondingly, the constraints are:

[0110] P cu (t)≤-P min +P(t)

[0111] P cd (t)≤P max -P(t)

[0112] Among them, P cu (t) is the upper limit reserve capacity, P cd(t) is the lower limit reserve capacity, P(t) is the operating power, P min is the minimum discharge power, P max is the maximum charging power;

[0113]

[0114] Among them, E min is the battery minimum charging power constraint, E base is the minimum power of electric vehicles, E exp is the expected power, P max is the maximum charging power, t dep is the departure time, t is the current time, t base is the base time;

[0115]

[0116] P min ≤P(t)≤P max

[0117] i c (t)+i d (t)≤1

[0118] Where, P(t) is the operating power, P c (t) is the charging power, P d (t) is the discharge power, i c (t) is the 0-1 charge-discharge state variable, i d (t) is the 0-1 charge-discharge state variable, i c (t) = 0 means the electric vehicle is not charged, i c (t) = 1 means the electric vehicle has been charged. For charging efficiency, is the discharge efficiency, P max is the maximum charging power, P min is the minimum discharge power;

[0119] The reserve capacity is analyzed and the objective function of the corresponding power linear programming is obtained by the following formula:

[0120]

[0121] M=M r.opt +M d.opt -M c.opt

[0122] M r.opt =P cu ×π cu

[0123] M d.opt =Pcd ×π cd

[0124] M c.opt =P d ×π d -P c ×π c

[0125]

[0126] Among them, M agg For aggregators to control the overall revenue of electric vehicles, M j is the total income of electric vehicles j participating in the aggregate regulation, M is the objective function, M r.opt is the reserve income of electric vehicles after participating in the aggregator dispatch, M d.opt is the emission income of electric vehicles after participating in the aggregator dispatch, M c.opt is the charging cost of electric vehicles after participating in the aggregator dispatch, π cu , π cd , π c , π d are the prices of upper reserve, lower reserve, charging power and discharging power respectively, N is the total number of electric vehicles in the cluster, The electric vehicle does not accept the charging and discharging power of centralized scheduling, P cu , P cd , P c , P d It is an intermediate variable related to power P.

[0127] Through the above scheme, this embodiment collects historical charging data and related behavior data of electric vehicles, cleans and normalizes the historical charging data and the behavior data, and obtains processed target data; uses a long short-term memory network LSTM to predict future behavior information of the electric vehicle based on the target data, and uses a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation based on the target data; trains and optimizes the LSTM-FNN model based on the future behavior information and the willingness information, calculates the spare capacity of the electric vehicle, analyzes the spare capacity, and obtains the objective function of the corresponding power linear programming, which can take into account the uncertainty of the user's willingness to participate, improves the accuracy and stability of the spare capacity prediction, and helps to improve the efficiency and reliability of the power system operation.

[0128] Based on the above hardware structure, an embodiment of a method for predicting the spare capacity of an electric vehicle of the present invention is proposed.

[0129] Reference Figure 2 , Figure 2It is a flow chart of the first embodiment of the method for predicting the spare capacity of an electric vehicle according to the present invention.

[0130] In a first embodiment, the electric vehicle reserve capacity prediction method comprises the following steps:

[0131] Step S10: Collect historical charging data and related behavior data of the electric vehicle, clean and normalize the historical charging data and the behavior data, and obtain processed target data.

[0132] It should be noted that after collecting the historical charging data and related behavior data of the electric vehicle, the historical charging data and the behavior data are cleaned and normalized to obtain the processed target data.

[0133] Step S20: using a long short-term memory network LSTM to predict future behavior information of the electric vehicle according to the target data, and using a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data.

[0134] It should be understood that the long short-term memory network (LSTM) is used to predict the future behavior information of the electric vehicle based on historical data, that is, the future behavior information of the electric vehicle is predicted according to the target data, and the fuzzy neural network (FNN) is used to analyze the willingness of the electric vehicle users to participate in centralized regulation, that is, the willingness information of the electric vehicle users to participate in centralized regulation is analyzed according to the target data.

[0135] Step S30: training and optimizing the LSTM-FNN model according to the future behavior information and the intention information, calculating and obtaining the spare capacity of the electric vehicle, analyzing the spare capacity, and obtaining the objective function of the corresponding power linear programming.

[0136] It can be understood that the LST M-FNN model is trained and optimized based on the future behavior information and the intention information, the spare capacity of the electric vehicle cluster is calculated, and the prediction results are analyzed by simulation tools to obtain the objective function of the corresponding power linear programming, thereby verifying the prediction accuracy and stability of the model.

[0137] Furthermore, the step S30 specifically includes the following steps:

[0138] The LSTM-FNN model is trained and optimized according to the future behavior information and the willingness information, and the spare capacity of the electric vehicle is obtained by the following calculation:

[0139]

[0140] Among them, P cu (k) is the upper limit reserve capacity, γ(k) is the adjustment coefficient, P min is the minimum discharge power, P(k) is the operating power, E(k) is the battery capacity of the electric vehicle, and E min (k+1) is the minimum power requirement for the next time step k+1, P cd (k) is the lower limit reserve capacity, P max is the maximum charging power, E max is the maximum battery capacity, Δt is the length of the time step;

[0141] Correspondingly, the constraints are:

[0142] P cu (t)≤-P min +P(t)

[0143] P cd (t)≤P max -P(t)

[0144] Among them, P cu (t) is the upper limit reserve capacity, P cd (t) is the lower limit reserve capacity, P(t) is the operating power, P min is the minimum discharge power, P max is the maximum charging power;

[0145]

[0146] Among them, E min is the battery minimum charging power constraint, E base is the minimum power of electric vehicles, E exp is the expected power, P max is the maximum charging power, t dep is the departure time, t is the current time, t base is the base time;

[0147]

[0148] P min ≤P(t)≤P max

[0149] i c (t)+i d (t)≤1

[0150] Where, P(t) is the operating power, P c (t) is the charging power, P d (t) is the discharge power, i c (t) is the 0-1 charge-discharge state variable, i d(t) is the 0-1 charge-discharge state variable, i c (t) = 0 means the electric vehicle is not charged, i c (t) = 1 means the electric vehicle has been charged. For charging efficiency, is the discharge efficiency, P max is the maximum charging power, P min is the minimum discharge power;

[0151] The reserve capacity is analyzed and the objective function of the corresponding power linear programming is obtained by the following formula:

[0152]

[0153] M=M r.opt +M d.opt -M c.opt

[0154] M r.opt =P cu ×π cu

[0155] M d.opt =P cd ×π cd

[0156] M c.opt =P d ×π d -P c ×π c

[0157]

[0158] Among them, M agg For aggregators to control the overall revenue of electric vehicles, M j is the total income of electric vehicles j participating in the aggregate regulation, M is the objective function, M r.opt is the reserve income of electric vehicles after participating in the aggregator dispatch, M d.opt is the emission income of electric vehicles after participating in the aggregator dispatch, M c.opt is the charging cost of electric vehicles after participating in the aggregator dispatch, π cu , π cd , π c , π d are the prices of upper reserve, lower reserve, charging power and discharging power respectively, N is the total number of electric vehicles in the cluster, P c 0 The electric vehicle does not accept the charging and discharging power of centralized scheduling, P cu , P cd , P c , Pd It is an intermediate variable related to power P.

[0159] In the specific implementation, t base The reference time is the time point when the power remains stable. When calculating the real-time power of electric vehicles, the battery power limitation must also be considered. The power of electric vehicles is lower than the guaranteed minimum power E base After that, the adjustment can be made. Constrained by the maximum charging power of the battery, the power of the electric vehicle must reach the expected power E exp At the same time, the real-time power of electric vehicles has a minimum constraint E during the grid connection time. min .

[0160] Through the above scheme, this embodiment collects historical charging data and related behavior data of electric vehicles, cleans and normalizes the historical charging data and the behavior data, and obtains processed target data; uses a long short-term memory network LSTM to predict future behavior information of the electric vehicle based on the target data, and uses a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation based on the target data; trains and optimizes the LSTM-FNN model based on the future behavior information and the willingness information, calculates the spare capacity of the electric vehicle, analyzes the spare capacity, and obtains the objective function of the corresponding power linear programming, which can take into account the uncertainty of the user's willingness to participate, improves the accuracy and stability of the spare capacity prediction, and helps to improve the efficiency and reliability of the power system operation.

[0161] Further, Figure 3 FIG. 1 is a flow chart of a second embodiment of a method for predicting the reserve capacity of an electric vehicle according to the present invention. Figure 3 As shown, a second embodiment of the electric vehicle reserve capacity prediction method of the present invention is proposed based on the first embodiment. In this embodiment, the step S10 specifically includes the following steps:

[0162] Step S11, collecting the arrival time, departure time, charging status, expected power and related behavior data of the electric vehicle in a historical time period, and taking the arrival time, departure time, charging status and expected power as historical charging data.

[0163] It should be noted that after collecting the arrival time, departure time, charging status, expected power and related behavior data of the electric vehicle in the historical time period, the arrival time, the departure time, the charging status (i.e., the state of charge (SOC)) and the expected power can be used as historical charging data.

[0164] Step S12: performing data cleaning on the historical charging data and the behavior data to remove missing values ​​and abnormal values ​​in the historical charging data and the behavior data.

[0165] It is understandable that data cleaning of the collected data, i.e., historical charging data and behavioral data, can eliminate missing values ​​and outliers.

[0166] Step S13: normalize the cleaned data to obtain processed target data.

[0167] It should be understood that the cleaned data can be normalized to obtain processed target data.

[0168] Through the above scheme, this embodiment collects the arrival time, departure time, charging status, expected power and related behavior data of the electric vehicle in the historical time period, and uses the arrival time, departure time, charging status and expected power as historical charging data; performs data cleaning on the historical charging data and the behavior data, and eliminates missing values ​​and abnormal values ​​in the historical charging data and the behavior data; and performs normalization on the cleaned data to obtain processed target data, which can take into account the uncertainty of user participation willingness and improve the accuracy and stability of spare capacity prediction.

[0169] Further, Figure 4 FIG. 1 is a flow chart of a third embodiment of a method for predicting the reserve capacity of an electric vehicle according to the present invention. Figure 4 As shown, a third embodiment of the electric vehicle reserve capacity prediction method of the present invention is proposed based on the first embodiment. In this embodiment, the step S20 specifically includes the following steps:

[0170] Step S21, determining a training set and a test set corresponding to the target data according to a preset ratio.

[0171] It should be noted that the training set and the test set corresponding to the target data can be determined according to a preset ratio.

[0172] In the specific implementation, due to the increase in sample depth, in order to avoid the gradient explosion problem, the LSTM network is used to predict the future behavior information of electric vehicles based on historical data, and the ratio of the training set to the test set is set to 8:2.

[0173] Step S22: using a long short-term memory network (LSTM) to predict and obtain future behavior information of the electric vehicle based on the training set and the test set.

[0174] It can be understood that the long short-term memory network LSTM is used to predict the future behavior information of the electric vehicle based on the training set and the test set.

[0175] Step S23, obtaining the grid connection time, SOC and electricity price from the target data, inputting the grid connection time, SOC and electricity price into the fuzzy neural network FNN, and outputting the willingness information of the user of the electric vehicle to participate in centralized regulation.

[0176] It should be understood that the grid connection time, SOC and electricity price can be obtained from the target data, and the grid connection time, SOC and electricity price are input into the fuzzy neural network FNN, which then outputs the willingness information of the user of the electric vehicle to participate in centralized regulation.

[0177] In the specific implementation, according to fuzzy reasoning, fuzzy data is extracted into the neural network, and the neural network continues to learn and extract fuzzy reasoning rules, so that the fuzzy neural network has generalization ability and can also process accurate information and fuzzy information at the same time. The three factors of grid connection time, SOC and electricity price are selected as the input of fuzzy reasoning, and the user's willingness to participate in the centralized regulation of the aggregator is used as the output.

[0178] Furthermore, the step S23 specifically includes the following steps:

[0179] The grid connection time is obtained from the target data by the following formula:

[0180] t=t step -t arr +t dep

[0181] Among them, t is the grid connection time, t step is the time step, representing a specific time, t arr is the arrival time of the electric vehicle, t dep Leave time for electric cars;

[0182] The state of charge SOC and the electricity price are obtained from the target data, the grid connection time, the SOC and the electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

[0183] It can be understood that by analyzing historical data, the membership functions of grid connection time, SOC and electricity price are obtained. According to the time scale of 1h, a day is divided into 24 time steps, that is, a total time step of 24 hours, which describes the time points when electric vehicles enter and leave charging; since the SOC interval itself is already between [0, 1], the predicted data can be directly selected as the membership value.

[0184] Furthermore, the step obtains the state of charge SOC and the electricity price from the target data, inputs the grid connection time, the SOC and the electricity price into the fuzzy neural network FNN, and outputs the willingness information of the user of the electric vehicle to participate in centralized regulation, specifically including the following steps:

[0185] Using charge prediction data from the target data as the state of charge SOC;

[0186] The income after the electric vehicle participates in the dispatch, the self-charging income, the reserve income, the emission income and the charging cost corresponding to the electricity price are obtained from the target data, and the profit loss of the electric vehicle user participating in the centralized regulation is calculated by the following formula based on the income after the dispatch, the self-charging income, the reserve income, the emission income and the charging cost:

[0187]

[0188] Among them, a is the profit loss, M L is the income after participating in the scheduling, M B is the self-charging income, M reserve For reserve income,

[0189] Determine the willingness information of the user of the electric vehicle to participate in the centralized regulation according to the profit loss, M c For charging cost.

[0190] In the specific implementation, the price factor uses the ratio of the self-charging cost to the cost after participating in the aggregator scheduling to describe the impact of the price factor, and the profit coefficient a is set to characterize the profit loss of users before and after the centralized regulation of the aggregator.

[0191] Through the above scheme, this embodiment determines the training set and the test set corresponding to the target data according to a preset ratio; uses the long short-term memory network LSTM to predict the future behavior information of the electric vehicle according to the training set and the test set; obtains the grid connection time, SOC and electricity price from the target data, inputs the grid connection time, the SOC and the electricity price into the fuzzy neural network FNN, and outputs the willingness information of the user of the electric vehicle to participate in centralized regulation, which can take into account the uncertainty of the user's willingness to participate and improve the accuracy and stability of the reserve capacity prediction.

[0192] Further, Figure 5 FIG. 4 is a flow chart of a fourth embodiment of a method for predicting the reserve capacity of an electric vehicle according to the present invention. Figure 5 As shown, a fourth embodiment of the electric vehicle reserve capacity prediction method of the present invention is proposed based on the third embodiment. In this embodiment, the step S22 specifically includes the following steps:

[0193] Step S221, input the training set and the test set into a long short-term memory network LSTM to obtain a network output result.

[0194] It should be noted that the training set and the test set can be input into a long short-term memory network LSTM, and then the network output result can be obtained.

[0195] Step S222: Calculate the mean square error between the network output result and the target value in the training set, and use the mean square error as the training loss function.

[0196] It can be understood that the mean square error between the network output result and the target value in the training set is calculated, and the mean square error can be used as the training loss function.

[0197] Step S223: Obtain an observed value from the network output result, and take the square root of the second-order sample moment of the difference between the predicted value of the test set and the observed value as the root mean square error.

[0198] It should be understood that the observed value is obtained from the network output result, and the square root of the second-order sample moment of the difference between the predicted value of the test set and the observed value can be used as the root mean square error.

[0199] Step S224: Prune the LSTM model according to the training loss function and the root mean square error to obtain a pruned target LSTM model.

[0200] It can be understood that the LSTM model can be pruned according to the training loss function and the root mean square error to obtain a pruned target LSTM model.

[0201] Step S225: predict the target data according to the target LSTM model to obtain future behavior information of the electric vehicle.

[0202] It should be understood that the target data can be predicted based on the target LSTM model, and the future behavior information of the electric vehicle can be obtained.

[0203] In the specific implementation, during the model training process, the training loss function and the root mean square error (RMSE) are visualized to prune it. The training loss function is the mean square error (MSE) between the network output and the target value in the training dataset; the root mean square error represents the square root of the second-order sample moment of the difference between the predicted value and the observed value (sample standard deviation), which is used to measure the difference between the model predicted value and the true value.

[0204] Through the above scheme, this embodiment obtains the network output result by inputting the training set and the test set into the long short-term memory network LSTM; calculates the mean square error between the network output result and the target value in the training set, and uses the mean square error as the training loss function; obtains the observation value from the network output result, and uses the square root of the second-order sample moment of the difference between the predicted value of the test set and the observed value as the root mean square error; prunes the LSTM model according to the training loss function and the root mean square error to obtain the pruned target LSTM model; predicts the target data according to the target LSTM model to obtain the future behavior information of the electric vehicle; and can take into account the uncertainty of user participation willingness, thereby improving the accuracy and stability of spare capacity prediction.

[0205] Accordingly, the present invention further provides a device for predicting the spare capacity of an electric vehicle.

[0206] Reference Figure 6 , Figure 6 This is a functional module diagram of the first embodiment of the electric vehicle spare capacity prediction device of the present invention.

[0207] In a first embodiment of the electric vehicle spare capacity prediction device of the present invention, the electric vehicle spare capacity prediction device comprises:

[0208] The data processing module 10 is used to collect the historical charging data and related behavior data of the electric vehicle, clean and normalize the historical charging data and the behavior data, and obtain processed target data.

[0209] The information analysis module 20 is used to predict the future behavior information of the electric vehicle according to the target data using a long short-term memory network LSTM, and to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data using a fuzzy neural network FNN.

[0210] The prediction and planning module 30 is used to train and optimize the LSTM-FNN model according to the future behavior information and the willingness information, calculate the spare capacity of the electric vehicle, analyze the spare capacity, and obtain the objective function of the corresponding power linear programming.

[0211] The data processing module 10 is also used to collect the arrival time, departure time, charging status, expected power and related behavior data of the electric vehicle within a historical time period, and use the arrival time, departure time, charging status and expected power as historical charging data; perform data cleaning on the historical charging data and the behavior data to eliminate missing values ​​and abnormal values ​​in the historical charging data and the behavior data; and perform normalization processing on the cleaned data to obtain processed target data.

[0212] The information analysis module 20 is further used to determine a training set and a test set corresponding to the target data according to a preset ratio; use a long short-term memory network LSTM to predict the future behavior information of the electric vehicle based on the training set and the test set; obtain the grid connection time, SOC and electricity price from the target data, input the grid connection time, the SOC and the electricity price into a fuzzy neural network FNN, and output the willingness information of the user of the electric vehicle to participate in centralized regulation.

[0213] The information analysis module 20 is also used to input the training set and the test set into a long short-term memory network LSTM to obtain a network output result; calculate the mean square error between the network output result and the target value in the training set, and use the mean square error as a training loss function; obtain an observation value from the network output result, and use the square root of the second-order sample moment of the difference between the predicted value of the test set and the observed value as the root mean square error; prune the LSTM model according to the training loss function and the root mean square error to obtain a pruned target LSTM model; predict the target data according to the target LSTM model to obtain future behavior information of the electric vehicle.

[0214] The information analysis module 20 is further used to obtain the grid connection time from the target data by the following formula:

[0215] t=t step -t arr +t dep

[0216] Among them, t is the grid connection time, t step is the time step, representing a specific time, t arr is the arrival time of the electric vehicle, t dep Leave time for electric cars;

[0217] The state of charge SOC and the electricity price are obtained from the target data, the grid connection time, the SOC and the electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

[0218] The information analysis module 20 is further used to use the charge prediction data in the target data as the state of charge SOC;

[0219] The income after the electric vehicle participates in the dispatch, the self-charging income, the reserve income, the emission income and the charging cost corresponding to the electricity price are obtained from the target data, and the profit loss of the electric vehicle user participating in the centralized regulation is calculated by the following formula based on the income after the dispatch, the self-charging income, the reserve income, the emission income and the charging cost:

[0220]

[0221] Among them, a is the profit loss, M L is the income after participating in the scheduling, M B is the self-charging income, M reserve For reserve income,

[0222] Determine the willingness information of the user of the electric vehicle to participate in the centralized regulation according to the profit loss, M c For charging cost.

[0223] The prediction planning module 30 is further used to train and optimize the LSTM-FNN model according to the future behavior information and the willingness information, and obtain the spare capacity of the electric vehicle by calculating the following formula:

[0224]

[0225] Among them, P cu (k) is the upper limit reserve capacity, γ(k) is the adjustment coefficient, P min is the minimum discharge power, P(k) is the operating power, E(k) is the battery capacity of the electric vehicle, and E min (k+1) is the minimum power requirement for the next time step k+1, P cd (k) is the lower limit reserve capacity, P max is the maximum charging power, E max is the maximum battery capacity, Δt is the length of the time step;

[0226] Correspondingly, the constraints are:

[0227] P cu (t)≤-P min +P(t)

[0228] P cd (t)≤P max -P(t)

[0229] Among them, P cu (t) is the upper limit reserve capacity, P cd (t) is the lower limit reserve capacity, P(t) is the operating power, P min is the minimum discharge power, P max is the maximum charging power;

[0230]

[0231] Among them, E min is the battery minimum charging power constraint, E base is the minimum power of electric vehicles, E exp is the expected power, Pmax is the maximum charging power, t dep is the departure time, t is the current time, t base is the base time;

[0232]

[0233] P min ≤P(t)≤P max

[0234] i c (t)+i d (t)≤1

[0235] Where, P(t) is the operating power, P c (t) is the charging power, P d (t) is the discharge power, i c (t) is the 0-1 charge-discharge state variable, i d (t) is the 0-1 charge-discharge state variable, i c (t) = 0 means the electric vehicle is not charged, i c (t) = 1 means the electric vehicle has been charged. For charging efficiency, is the discharge efficiency, P max is the maximum charging power, P min is the minimum discharge power;

[0236] The reserve capacity is analyzed and the objective function of the corresponding power linear programming is obtained by the following formula:

[0237]

[0238] M=M r.opt +M d.opt -M c.opt

[0239] M r.opt =P cu ×π cu

[0240] M d.opt =P cd ×π cd

[0241] M c.opt =P d ×π d -P c ×π c

[0242]

[0243] Among them, Magg For aggregators to control the overall revenue of electric vehicles, M j is the total income of electric vehicles j participating in the aggregate regulation, M is the objective function, M r.opt is the reserve income of electric vehicles after participating in the aggregator dispatch, M d.opt is the emission income of electric vehicles after participating in the aggregator dispatch, M c.opt is the charging cost of electric vehicles after participating in the aggregator dispatch, π cu , π cd , π c , π d are the prices of upper reserve, lower reserve, charging power and discharging power respectively, N is the total number of electric vehicles in the cluster, The electric vehicle does not accept the charging and discharging power of centralized scheduling, P cu , P cd , P c , P d It is an intermediate variable related to power P.

[0244] The steps implemented by each functional module of the electric vehicle spare capacity prediction device can refer to the various embodiments of the electric vehicle spare capacity prediction method of the present invention, and will not be repeated here.

[0245] In addition, an embodiment of the present invention further provides a storage medium, on which an electric vehicle spare capacity prediction program is stored. When the electric vehicle spare capacity prediction program is executed by a processor, the following operations are implemented:

[0246] Collecting historical charging data and related behavior data of electric vehicles, cleaning and normalizing the historical charging data and the behavior data, and obtaining processed target data;

[0247] Using a long short-term memory network LSTM to predict future behavior information of the electric vehicle according to the target data, and using a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data;

[0248] The LSTM-FNN model is trained and optimized according to the future behavior information and the intention information, the spare capacity of the electric vehicle is calculated, the spare capacity is analyzed, and the objective function of the corresponding power linear programming is obtained.

[0249] Furthermore, when the electric vehicle reserve capacity prediction program is executed by the processor, the following operations are also implemented:

[0250] Collecting the arrival time, departure time, charging state, expected power and related behavior data of the electric vehicle in a historical time period, and using the arrival time, departure time, charging state and expected power as historical charging data;

[0251] Performing data cleaning on the historical charging data and the behavior data to remove missing values ​​and abnormal values ​​in the historical charging data and the behavior data;

[0252] The cleaned data is normalized to obtain the processed target data.

[0253] Furthermore, when the electric vehicle reserve capacity prediction program is executed by the processor, the following operations are also implemented:

[0254] Determine a training set and a test set corresponding to the target data according to a preset ratio;

[0255] Using a long short-term memory network (LSTM) to predict and obtain future behavior information of the electric vehicle based on the training set and the test set;

[0256] The grid connection time, SOC and electricity price are obtained from the target data, the grid connection time, SOC and electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

[0257] Furthermore, when the electric vehicle reserve capacity prediction program is executed by the processor, the following operations are also implemented:

[0258] Input the training set and the test set into a long short-term memory network LSTM to obtain a network output result;

[0259] Calculate the mean square error between the network output result and the target value in the training set, and use the mean square error as the training loss function;

[0260] Obtain an observed value from the network output result, and take the square root of the second-order sample moment of the difference between the predicted value of the test set and the observed value as the root mean square error;

[0261] Pruning the LSTM model according to the training loss function and the root mean square error to obtain a pruned target LSTM model;

[0262] The target data is predicted according to the target LSTM model to obtain future behavior information of the electric vehicle.

[0263] Furthermore, when the electric vehicle reserve capacity prediction program is executed by the processor, the following operations are also implemented:

[0264] The grid connection time is obtained from the target data by the following formula:

[0265] t=t step -t arr +t dep

[0266] Among them, t is the grid connection time, t step is the time step, representing a specific time, t arr is the arrival time of the electric vehicle, t dep Leave time for electric cars;

[0267] The state of charge SOC and the electricity price are obtained from the target data, the grid connection time, the SOC and the electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

[0268] Furthermore, when the electric vehicle reserve capacity prediction program is executed by the processor, the following operations are also implemented:

[0269] Using charge prediction data from the target data as the state of charge SOC;

[0270] The income after the electric vehicle participates in the dispatch, the self-charging income, the reserve income, the emission income and the charging cost corresponding to the electricity price are obtained from the target data, and the profit loss of the electric vehicle user participating in the centralized regulation is calculated by the following formula based on the income after the dispatch, the self-charging income, the reserve income, the emission income and the charging cost:

[0271]

[0272] Among them, a is the profit loss, M L is the income after participating in the scheduling, M B is the self-charging income, M reserve For reserve income,

[0273] Determine the willingness information of the user of the electric vehicle to participate in the centralized regulation according to the profit loss, M c For charging cost.

[0274] Furthermore, when the electric vehicle reserve capacity prediction program is executed by the processor, the following operations are also implemented:

[0275] The LSTM-FNN model is trained and optimized according to the future behavior information and the willingness information, and the spare capacity of the electric vehicle is obtained by the following calculation:

[0276]

[0277] Among them, P cu(k) is the upper limit reserve capacity, γ(k) is the adjustment coefficient, P min is the minimum discharge power, P(k) is the operating power, E(k) is the battery capacity of the electric vehicle, and E min (k+1) is the minimum power requirement for the next time step k+1, P cd (k) is the lower limit reserve capacity, P max is the maximum charging power, E max is the maximum battery capacity, Δt is the length of the time step;

[0278] Correspondingly, the constraints are:

[0279] P cu (t)≤-P min +P(t)

[0280] P cd (t)≤P max -P(t)

[0281] Among them, P cu (t) is the upper limit reserve capacity, P cd (t) is the lower limit reserve capacity, P(t) is the operating power, P min is the minimum discharge power, P max is the maximum charging power;

[0282]

[0283] Among them, E min is the battery minimum charging power constraint, E base is the minimum power of electric vehicles, E exp is the expected power, P max is the maximum charging power, t dep is the departure time, t is the current time, t base is the base time;

[0284]

[0285] P min ≤P(t)≤P max

[0286] i c (t)+i d (t)≤1

[0287] Where, P(t) is the operating power, P c (t) is the charging power, P d (t) is the discharge power, i c (t) is the 0-1 charge-discharge state variable, i d (t) is the 0-1 charge-discharge state variable, ic (t) = 0 means the electric vehicle is not charged, i c (t) = 1 means the electric vehicle has been charged. For charging efficiency, is the discharge efficiency, P max is the maximum charging power, P min is the minimum discharge power;

[0288] The reserve capacity is analyzed and the objective function of the corresponding power linear programming is obtained by the following formula:

[0289]

[0290] M=M r.opt +M d.opt -M c.opt

[0291] M r.opt =P cu ×π cu

[0292] M d.opt =P cd ×π cd

[0293] M c.opt =P d ×π d -P c ×π c

[0294]

[0295] Among them, M agg For aggregators to control the overall revenue of electric vehicles, M j is the total income of electric vehicles j participating in the aggregate regulation, M is the objective function, M r.opt is the reserve income of electric vehicles after participating in the aggregator dispatch, M d.opt is the emission income of electric vehicles after participating in the aggregator dispatch, M c.opt is the charging cost of electric vehicles after participating in the aggregator dispatch, π cu , π cd , π c , π d are the prices of upper reserve, lower reserve, charging power and discharging power respectively, N is the total number of electric vehicles in the cluster, The electric vehicle does not accept the charging and discharging power of centralized scheduling, P cu , P cd , P c , P d It is an intermediate variable related to power P.

[0296] Those skilled in the art will appreciate that all or part of the steps in the above-mentioned implementation method can be accomplished by instructing related hardware through a program, and the program is stored in a storage medium, including a number of instructions for enabling a device (which may be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the method described in each implementation method of the present application; and the aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0297] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0298] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0299] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for predicting the spare capacity of an electric vehicle, characterized in that: The electric vehicle reserve capacity prediction method comprises: Collecting historical charging data and related behavior data of electric vehicles, cleaning and normalizing the historical charging data and the behavior data, and obtaining processed target data; Using a long short-term memory network LSTM to predict future behavior information of the electric vehicle according to the target data, and using a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data; The LSTM-FNN model is trained and optimized according to the future behavior information and the intention information, the spare capacity of the electric vehicle is calculated, the spare capacity is analyzed, and the objective function of the corresponding power linear programming is obtained.

2. The electric vehicle reserve capacity prediction method according to claim 1, characterized in that: The collecting of historical charging data and related behavior data of the electric vehicle, cleaning and normalizing the historical charging data and the behavior data, and obtaining processed target data include: Collecting the arrival time, departure time, charging state, expected power and related behavior data of the electric vehicle in a historical time period, and using the arrival time, departure time, charging state and expected power as historical charging data; Performing data cleaning on the historical charging data and the behavior data to remove missing values ​​and abnormal values ​​in the historical charging data and the behavior data; The cleaned data is normalized to obtain the processed target data.

3. The electric vehicle reserve capacity prediction method according to claim 1, characterized in that: The method of using a long short-term memory network LSTM to predict future behavior information of the electric vehicle according to the target data, and using a fuzzy neural network FNN to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data, includes: Determine a training set and a test set corresponding to the target data according to a preset ratio; Using a long short-term memory network (LSTM) to predict and obtain future behavior information of the electric vehicle based on the training set and the test set; The grid connection time, SOC and electricity price are obtained from the target data, the grid connection time, SOC and electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

4. The electric vehicle reserve capacity prediction method according to claim 3, characterized in that: The method of using a long short-term memory network (LSTM) to predict and obtain future behavior information of the electric vehicle according to the training set and the test set includes: Input the training set and the test set into a long short-term memory network LSTM to obtain a network output result; Calculate the mean square error between the network output result and the target value in the training set, and use the mean square error as the training loss function; Obtain an observed value from the network output result, and take the square root of the second-order sample moment of the difference between the predicted value of the test set and the observed value as the root mean square error; Pruning the LSTM model according to the training loss function and the root mean square error to obtain a pruned target LSTM model; The target data is predicted according to the target LSTM model to obtain future behavior information of the electric vehicle.

5. The electric vehicle reserve capacity prediction method according to claim 3, characterized in that: The grid connection time, SOC and electricity price are obtained from the target data, the grid connection time, SOC and electricity price are input into the fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in the centralized control is output, including: The grid connection time is obtained from the target data by the following formula: t=t step -t arr +t dep Among them, t is the grid connection time, t step is the time step, representing a specific time, t arr is the arrival time of the electric vehicle, t dep Leave time for electric cars; The state of charge SOC and the electricity price are obtained from the target data, the grid connection time, the SOC and the electricity price are input into a fuzzy neural network FNN, and the willingness information of the user of the electric vehicle to participate in centralized regulation is output.

6. The electric vehicle reserve capacity prediction method according to claim 5, characterized in that: The method of obtaining the state of charge SOC and the electricity price from the target data, inputting the grid connection time, the SOC and the electricity price into a fuzzy neural network FNN, and outputting the willingness information of the user of the electric vehicle to participate in centralized regulation includes: Using charge prediction data from the target data as the state of charge SOC; The income after the electric vehicle participates in the dispatch, the self-charging income, the reserve income, the emission income and the charging cost corresponding to the electricity price are obtained from the target data, and the profit loss of the electric vehicle user participating in the centralized regulation is calculated by the following formula based on the income after the dispatch, the self-charging income, the reserve income, the emission income and the charging cost: Among them, a is the profit loss, M L is the income after participating in the scheduling, M B is the self-charging income, M reserve For reserve income, Determine the willingness information of the user of the electric vehicle to participate in the centralized regulation according to the profit loss, M c For charging cost.

7. The electric vehicle reserve capacity prediction method according to claim 1, characterized in that: The step of training and optimizing the LSTM-FNN model according to the future behavior information and the willingness information, calculating and obtaining the spare capacity of the electric vehicle, analyzing the spare capacity, and obtaining the objective function of the corresponding power linear programming includes: The LSTM-FNN model is trained and optimized according to the future behavior information and the willingness information, and the spare capacity of the electric vehicle is obtained by the following calculation: Among them, P cu (k) is the upper limit reserve capacity, γ(k) is the adjustment coefficient, P min is the minimum discharge power, P(k) is the operating power, E(k) is the battery capacity of the electric vehicle, and E min (k+1) is the minimum power requirement for the next time step k+1, P cd (k) is the lower limit spare capacity, P max is the maximum charging power, E max is the maximum battery capacity, Δt is the length of the time step; Correspondingly, the constraints are: P cu (t)≤-P min +P(t) P cd (t)≤P max -P(t) Among them, P cu (t) is the upper limit reserve capacity, P cd (t) is the lower limit reserve capacity, P(t) is the operating power, P min is the minimum discharge power, P max is the maximum charging power; Among them, E min is the battery minimum charging power constraint, E base is the minimum power of electric vehicles, E exp is the expected power, P max is the maximum charging power, t dep is the departure time, t is the current time, t base is the base time; P min ≤P(t)≤P max i c (t)+i d (t)≤1 Where, P(t) is the operating power, P c (t) is the charging power, P d (t) is the discharge power, i c (t) is the 0-1 charge-discharge state variable, i d (t) is the 0-1 charge-discharge state variable, i c (t) = 0 means the electric vehicle is not charged, i c (t) = 1 means the electric vehicle has been charged. For charging efficiency, is the discharge efficiency, P max is the maximum charging power, P min is the minimum discharge power; The reserve capacity is analyzed and the objective function of the corresponding power linear programming is obtained by the following formula: M=M r.opt +M d.opt -M c.opt M r.opt =P cu ×π cu M d.opt =P cd ×π cd M c.opt =P d ×π d -P c ×π c Among them, M agg For aggregators to control the overall revenue of electric vehicles, M j is the total income of electric vehicles j participating in the aggregate regulation, M is the objective function, M r.opt is the reserve income of electric vehicles after participating in the aggregator dispatch, M d.opt is the emission income of electric vehicles after participating in the aggregator dispatch, M c.opt is the charging cost of electric vehicles after participating in the aggregator dispatch, π cu , π cd , π c , π d are the prices of upper reserve, lower reserve, charging power and discharging power respectively, N is the total number of electric vehicles in the cluster, The electric vehicle does not accept the charging and discharging power of centralized scheduling, P cu , P cd , P c , P d It is an intermediate variable related to power P.

8. An electric vehicle spare capacity prediction device, characterized in that: The electric vehicle reserve capacity prediction device comprises: A data processing module, used to collect historical charging data and related behavior data of the electric vehicle, clean and normalize the historical charging data and the behavior data, and obtain processed target data; An information analysis module, used to predict the future behavior information of the electric vehicle according to the target data using a long short-term memory network LSTM, and to analyze the willingness information of the user of the electric vehicle to participate in centralized regulation according to the target data using a fuzzy neural network FNN; The prediction and planning module is used to train and optimize the LSTM-FNN model according to the future behavior information and the willingness information, calculate the spare capacity of the electric vehicle, analyze the spare capacity, and obtain the objective function of the corresponding power linear programming.

9. An electric vehicle spare capacity prediction device, characterized in that: The electric vehicle reserve capacity prediction device comprises: a memory, a processor and an electric vehicle reserve capacity prediction program stored in the memory and executable on the processor, wherein the electric vehicle reserve capacity prediction program is configured to implement the steps of the electric vehicle reserve capacity prediction method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium stores an electric vehicle spare capacity prediction program, which, when executed by a processor, implements the steps of the electric vehicle spare capacity prediction method according to any one of claims 1 to 7.

Images