Electric vehicle charging scheduling method, device and equipment and storage medium

Abstract

The invention provides an electric vehicle charging scheduling method. The method comprises the following steps: acquiring driving track data of a plurality of vehicles, and constructing a grid heat matrix according to the data; a semantic space is generated through singular value decomposition, and the space implies the degree of interest of a driver on each grid. Then, clustering the driving track data to generate driving track cluster data; and establishing a battery charging and discharging model with the goal of minimizing the difference between the charging cost and the discharging income, screening variable parameters by using a Monte Carlo method, and determining an optimal combination. And finally, calculating an electric vehicle charging scheduling scheme based on the optimal combination. According to the method, through driving track data aggregation and charging and discharging model calculation, the problem that acquisition of mass data and the precision of a battery charging model are difficult to meet at the same time is avoided.

Description

Technical Field

[0001] This application belongs to the field of electric vehicle charging scheduling, and particularly relates to an electric vehicle charging scheduling method, device, equipment and storage medium. Background Art

[0002] The scheduling decision of electric vehicles requires a large amount of driving trajectory data. However, in some real scenarios, it is a challenge to obtain sufficient data. In addition, for the needs of actual projects, the battery charging model often needs to simplify equations. However, this simplification requires a large number of parameters to complete the calculation, inevitably introducing errors. Therefore, in the scheduling decision of electric vehicles, it is difficult to simultaneously meet the data acquisition and the accuracy of the battery charging model. Summary of the Invention

[0003] The purpose of this application is to overcome the defects in the prior art and provide an electric vehicle charging scheduling method, device, equipment and storage medium.

[0004] This application provides an electric vehicle charging scheduling method, including:

[0005] Define the area to be scheduled as a scheduling space and divide it into multiple grids according to longitude and latitude;

[0006] Collect the driving trajectory data composed of longitude, latitude and time of multiple vehicles in the scheduling space;

[0007] Construct a grid heat matrix according to the driving trajectory data and the grids;

[0008] Perform singular value decomposition on the grid heat matrix to generate a semantic space composed of several left singular orthogonal vectors, and the semantic space implies the degree of interest of drivers in each grid;

[0009] Cluster the driving trajectory data according to the semantic space to generate driving trajectory cluster data;

[0010] Establish a battery charge and discharge model for electric vehicles with the goal of minimizing the difference between charging costs and discharge benefits;

[0011] Use the Monte Carlo method to screen the variable parameters in the battery charge and discharge model with the driving trajectory cluster data as the initial variable parameter combination, and determine the optimal variable parameter combination;

[0012] Based on the battery charge and discharge model with the optimal variable parameter combination, calculate the electric vehicle charging scheduling plan.

[0013] Optionally, constructing a grid heat matrix according to the driving trajectory data and the grids includes:

[0014] Construct a grid-vehicle matrix based on the driving trajectory data and the grid. Each element in the grid-vehicle matrix represents the residence time of the vehicle in the corresponding grid;

[0015] Record the elements with residence time less than the preset time length as 0, and convert the grid-vehicle matrix into a grid heat matrix.

[0016] Optionally, the degree of interest is represented by the inner product of the eigenvector after singular value decomposition and the corresponding singular value.

[0017] Optionally, the battery charge and discharge model includes:

[0018]

[0019]

[0020]

[0021] Wherein, is the charging cost, is the discharge income, is the battery loss cost, and are respectively the charging cost and discharge income of the electric vehicle after reaching the destination in the kth trip, and are binary variables. When their values are 1, they respectively represent that the electric vehicle charges or discharges at time t after reaching the destination in the kth trip, and are respectively the charging electricity price and discharge electricity price at time t, is the charge and discharge power of the electric vehicle at the destination of the kth trip C b is the battery replacement price, N b is the maximum charge and discharge times of the battery, N ch and N di are respectively the charging times and discharge times of the battery in a day, η ch and η di are respectively the charging efficiency and discharge efficiency.

[0022] Optionally, use the Monte Carlo method to screen the variable parameters in the battery charge and discharge model. The determination steps of the iteration times include:

[0023] Generate fitness according to the initial variable parameter combination;

[0024] According to the fitness, evolutionary computation is performed using an evolutionary algorithm, including: defining the number of ineffective evolutions and initializing it to 0; after each evolution of the evolutionary algorithm is completed, a new fitness is generated. When the absolute value of the difference between the nth fitness and the (n - 1)th fitness is less than the preset evolution accuracy, the number of ineffective evolutions is incremented by 1; when the absolute value of the difference between the nth fitness and the (n - 1)th fitness is greater than or equal to the preset evolution accuracy, the number of ineffective evolutions is cleared and the evolutionary computation is stopped.

[0025] Optionally, the Monte Carlo method is used to screen the variable parameters in the scheduling model, including:

[0026] Model the uncertainty of the variable parameter combinations of the battery charge and discharge model, and use Bayes' formula to select the variable parameter combination that is most consistent with the observed values to update the uncertainty of the next iteration;

[0027] When the iteration is completed, the variable parameter combination after iteration is obtained to complete the screening.

[0028] Optionally, perform singular value decomposition on the grid heat matrix to generate a semantic space composed of several left singular orthogonal vectors, including:

[0029] Perform singular value decomposition on the grid heat matrix according to the singular value decomposition formula E = M·diag(λ1, λ2, …, λs)·N';

[0030] where λ i is the singular value of matrix E; the eigenvectors of EE’ are the column vectors of M; the eigenvectors of E’E are the column vectors of N; s is the rank of matrix E;

[0031] After obtaining the column vectors of M, use the left singular vectors, that is, the column vectors of M, to generate the semantic space, including: for each left singular vector M_i, calculate its inner product with the corresponding singular value σ_i.

[0032] This application also provides an electric vehicle charging scheduling device, including:

[0033] A grid module, used to define the area to be scheduled as a scheduling space and divide it into multiple grids according to longitude and latitude;

[0034] An acquisition module, used to acquire the driving trajectory data composed of the longitude, latitude and time of multiple vehicles in the scheduling space;

[0035] A matrix module, used to construct a grid heat matrix according to the driving trajectory data and the grid;

[0036] A decomposition module, configured to perform singular value decomposition on the grid heat matrix to generate a semantic space composed of a plurality of left singular orthogonal vectors, where the semantic space implies the degree of interest of the driver in each grid;

[0037] A clustering module, configured to cluster the driving trajectory data according to the semantic space to generate driving trajectory cluster data;

[0038] A model module, configured to establish a battery charge and discharge model for an electric vehicle with the goal of minimizing the difference between the charging cost and the discharging income;

[0039] A screening module, configured to use the Monte Carlo method to screen the variable parameters in the battery charge and discharge model with the driving trajectory cluster data as the initial variable parameter combination to determine the optimal variable parameter combination;

[0040] A calculation module, configured to calculate an electric vehicle charging scheduling scheme based on the battery charge and discharge model with the optimal variable parameter combination.

[0041] This application also provides an electric vehicle charging scheduling device, including:

[0042] A memory, configured to store a computer executable program for the above-mentioned electric vehicle charging scheduling method;

[0043] A processor, configured to retrieve the computer executable program from the memory and execute: defining the area to be scheduled as a scheduling space and dividing it into multiple grids according to longitude and latitude; collecting driving trajectory data composed of the longitude, latitude, and time of multiple vehicles in the scheduling space; constructing a grid heat matrix based on the driving trajectory data and the grids; performing singular value decomposition on the grid heat matrix to generate a semantic space composed of a plurality of left singular orthogonal vectors, where the semantic space implies the degree of interest of the driver in each grid; clustering the driving trajectory data according to the semantic space to generate driving trajectory cluster data; establishing a battery charge and discharge model for an electric vehicle with the goal of minimizing the difference between the charging cost and the discharging income; using the Monte Carlo method to screen the variable parameters in the battery charge and discharge model with the driving trajectory cluster data as the initial variable parameter combination to determine the optimal variable parameter combination; calculating an electric vehicle charging scheduling scheme based on the battery charge and discharge model with the optimal variable parameter combination.

[0044] This application also provides a storage medium, storing a computer executable program, which is used to be retrieved by a processor to execute the steps of the above-mentioned electric vehicle charging scheduling method.

[0045] Advantages and beneficial effects of this application:

[0046] The present application provides an electric vehicle charging scheduling method, including: collecting driving trajectory data composed of the longitude, latitude, and time of multiple vehicles within the scheduling space; constructing a grid heat matrix based on the driving trajectory data and the grid; performing singular value decomposition on the grid heat matrix to generate a semantic space composed of several left singular orthogonal vectors, where the semantic space implies the degree of interest of drivers in each grid; clustering the driving trajectory data according to the semantic space to generate driving trajectory cluster data; establishing a battery charge and discharge model for electric vehicles with the goal of minimizing the difference between the charging cost and the discharge revenue; using the Monte Carlo method to screen the variable parameters in the battery charge and discharge model with the driving trajectory cluster data as the initial variable parameter combination to determine the optimal variable parameter combination; and calculating an electric vehicle charging scheduling plan based on the battery charge and discharge model with the optimal variable parameter combination. Through the aggregation of driving trajectory data in the present application and then combined with the charge and discharge model for calculation, it is not necessary to obtain a large amount of data, avoiding the situation where it is difficult to simultaneously meet the acquisition of a large amount of data and the accuracy of the battery charging model. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a schematic diagram of the electric vehicle charging scheduling process in the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The following further describes the present application with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand the present application and be able to implement it.

[0049] The following content is all examples of the specific implementation process provided to detail the technical solution to be protected by the present application. However, the present application can also be implemented in other ways different from the descriptions herein. Those skilled in the art can implement the present application using different technical means under the guidance of the concept of the present application. Therefore, the present application is not limited by the following specific embodiments.

[0050] As Figure 1 shown, the steps of the electric vehicle charging scheduling method in the present application include:

[0051] S101 Define the area to be scheduled as the scheduling space and divide it into multiple grids according to longitude and latitude.

[0052] Divide the area to be scheduled according to longitude and latitude. The longitude range is [L o,min , L o,max , and the latitude range is [L a,min , L a,max .

[0053] Then, according to these two ranges, represent the scheduling space X as an integral form of space, that is:

[0054]

[0055] This formula indicates that the entire scheduling space X is composed of an area where the latitude is between [L a,max , L a,min , and the longitude is between [L o,max , L o,min .

[0056] The scheduling space X is partitioned into M rows and N columns according to the latitude - first principle, and the formed (M×N) grids are called the grid space Y. Each grid is called a spatial grid y i , that is:

[0057]

[0058] The relationship between the scheduling space X and the grid space Y is X≡Y, and for any i, j (i≠j), it satisfies y i ∩y j =Φ.

[0059] In the scheduling area, for any given position x(g, h), a unique y i can be found in the grid space Y such that x∈y i .

[0060] S102 collects the driving trajectory data composed of the longitude, latitude, and time of multiple vehicles within the said scheduling space.

[0061] The space, time, and other information of the vehicle during driving are collected through the Beidou system, and the formed time series is called the driving trajectory R i ={R1, R2, …R k}, where R i =(o i , a i , t i ) is the spatio - temporal data containing longitude, latitude, and time information; k = len(R i ) is the length of the driving trajectory time series.

[0062] S103 constructs a grid heat matrix based on the said driving trajectory data and the said grid.

[0063] Suppose there are p vehicles accessing the trajectory data platform. Each vehicle is a column of the matrix, and the number of grids in the scheduling space is q. Each grid is a row of the matrix, forming a q×p "grid - vehicle" matrix F, which is expressed as follows:

[0064]

[0065] The elements of matrix F are assigned as the effective number of times the vehicle stays in the corresponding grid for more than 0.5h (staying less than 0.5h cannot be charged and is recorded as 0), and the grid heat matrix E can be obtained, which is expressed as:

[0066]

[0067] S104 performs singular value decomposition on the grid heat matrix to generate a semantic space composed of several left singular orthogonal vectors, and the semantic space implies the degree of interest of the driver in each grid.

[0068] Performing singular value decomposition on matrix E generates a semantic space composed of several left singular orthogonal vectors. This space implies the interest situation of the driver in each location, and the degree of interest is obtained from the inner product of the eigenvector of the space and the corresponding singular value.

[0069] Specifically, according to the singular value decomposition theorem, matrix E is an m×n-order matrix, and there exist n-order orthogonal matrices M = {m1, m2, …, m s} and N = {n1, n2, …, n s} such that:

[0070] E = M·diag(λ1, λ2, …, λ s )·N';

[0071] Among them, λ i is the singular value of matrix E; the eigenvectors of EE’ are the column vectors of M; the eigenvectors of E’E are the column vectors of N; s is the rank of matrix E.

[0072] After obtaining the column vectors of M, use the left singular vectors, that is, the column vectors of M, to generate the semantic space, including: for each left singular vector M_i, calculate its inner product with the corresponding singular value σ_i.

[0073] S105 clusters the driving trajectory data according to the semantic space to generate driving trajectory cluster data.

[0074] Performing singular value decomposition on matrix E to obtain the left singular vector matrix, which reflects the information space of the context semantics between spatial grids. What reflects the original signal is the first l component signals of its semantic information space (the size of l is determined by the singular values after specific matrix decomposition), which is expressed as:

[0075] El = Ml·diag(λ1, λ2, …, λl)·Nl';

[0076] The left singular vector M l is the dimensionality-reduced semantic subspace. The selection of the dimension l affects the performance of the algorithm. If the (l + 1)-th singular value decreases significantly slower compared with the previous one, it means that the value of l is optional.

[0077] In addition, within the dimensionality-reduced subspace, the semantic space mainly represents the common information among grids. Therefore, the weighted semantic subspace can be expressed as the product of the column vectors of the dimensionality-reduced semantic subspace and their corresponding singular values, expressed as:

[0078]

[0079] where each of the dimensionality-reduced subspaces is a class.

[0080] In this semantic space, the l value is considered a distance metric for measuring the correlation between each spatial grid. The correlation is obtained by calculating the distance between two grids, and the closer the distance, the higher the correlation between the two grids.

[0081] Using this distance metric, cluster the spatial grids. Specifically, select several grids with the closest distance and group them into one cluster. In this way, the entire spatial grid is divided into several clusters, and each cluster represents an area that a driver may be interested in.

[0082] When performing clustering, divide the semantic space into several regions according to the driving trajectory of the electric vehicle, and each region corresponds to a path that a driver may pass through. In this way, the driver's behavior can be predicted more accurately, improving the accuracy of clustering.

[0083] S106 establishes a battery charge-discharge model for an electric vehicle with the goal of minimizing the difference between the charging cost and the discharging revenue.

[0084] On the premise of meeting the travel demand, establish an optimization model for the battery charge and discharge of an electric vehicle:

[0085]

[0086]

[0087]

[0088] where F represents the battery charge-discharge model, is the charging cost, is the discharging revenue, is the battery loss cost, and are respectively the charging cost and the discharging revenue of the electric vehicle after reaching the destination in the k-th trip, and are binary variables, and when their values are 1, they respectively indicate that the electric vehicle charges or discharges at time t after reaching the destination in the k-th trip, and are respectively the charging electricity price and the discharging electricity price at time t, The charging and discharging power for the electric vehicle at the destination of the k-th trip , C b is the battery replacement price, N b is the maximum number of charge and discharge cycles of the battery, N ch and N di are the number of charging times and discharging times of the battery in a day respectively, η ch and η di are the charging efficiency and discharging efficiency respectively.

[0089] At any moment t, the electric vehicle can only be in one of the states of charging, discharging, and driving, so it is necessary to satisfy:

[0090]

[0091] To alleviate range anxiety, at the start of the trip, or at the end of only charging or only discharging, the battery charge should be greater than the minimum charging threshold 0.3Db and the power consumption of the next trip, and not exceed the maximum battery charge. Therefore, the following constraints need to be satisfied:

[0092]

[0093]

[0094]

[0095] In the formula, and are the charging power and discharging power of the electric vehicle after reaching the destination of the k-th trip respectively, D b is the maximum battery charge, D0 is the initial charge of the electric vehicle in a day before optimization. To make the charging load before and after optimization more comparable, the deviation of the battery charge at the end of the trip should not exceed 5% compared with the termination charge before optimization, that is:

[0096]

[0097] In the formula, D T is the termination charge of the electric vehicle in a day before optimization.

[0098] S107 uses the driving trajectory cluster data as the initial variable parameter combination, and uses the Monte Carlo method to screen the variable parameters in the battery charge and discharge model to determine the optimal variable parameter combination.

[0099] S1071 Based on the variable parameters of the battery charge and discharge model, select the driving trajectory cluster data as the initial parameter combination and set the initial uncertainty. This uncertainty is an estimated value or is determined based on prior knowledge or experience.

[0100] S1072 Sample using the Monte Carlo method according to the set parameter combination. Each sampling point represents a possible solution, that is, a possible data generation combination.

[0101] S1073 Evaluate each sampling point and calculate the similarity between the data it generates and the actual data. This evaluation is based on the accuracy of model prediction or other suitable similarity measurement methods.

[0102] S1074 Use Bayes' formula to gradually update the corresponding uncertainty according to the evaluation results of each sampling point. This update process is repeated as needed, and new observed data is used to update the uncertainty each time.

[0103] S1075 Select a new parameter combination and repeat steps S1072 - S1074 until the preset number of iterations is reached or other stopping conditions are met.

[0104] In this application, until the preset number of iterations is reached, it includes:

[0105] Based on the evolutionary number adaptive improvement method, add an "invalid evolutionary number tracking" process on the basis of the above process, so that the evolutionary number changes adaptively with the different dimensions of the search space.

[0106] Define the invalid evolutionary number as Nu and initialize it to 0. After each evolution of the PSO algorithm, a fitness fness,j will be generated. When the absolute value of the difference between this fitness and the fitness generated after the previous evolution is less than the preset evolution accuracy δp for the fusion of the driving trajectory data and the battery charge and discharge model, the invalid evolutionary number is incremented by 1; when the absolute value of the difference between the fitness generated after each evolution and the fitness generated after the previous evolution is greater than or equal to the evolution accuracy δp, the invalid evolutionary number is cleared, which is expressed as

[0107]

[0108] At this time, the evolution ends, that is, the preset number of iterations is reached.

[0109] S108 Calculate the charging and discharging states of the batteries in each grid based on the scheduling model of the optimal variables to obtain an electric vehicle charging scheduling plan.

[0110] With the goal of minimizing the operating cost of the microgrid and maximizing the economic benefits, that is:

[0111]

[0112] The objective function contains 4 parts, namely the power generation cost of the diesel generator in period t The battery loss cost of the electric vehicle Demand response load regulation cost Purchase and sale electricity interaction cost

[0113] Power balance constraint:

[0114]

[0115] Among them, Pi is the output power of distributed generation unit i; P gird is the electric energy provided by the superior power grid; P load is the sum of the power of each electrical load; P EV is the sum of the power of the electric vehicle cluster in the microgrid.

[0116] When the value of P EV is positive, it means that the electric vehicle consumes the electric energy of the power grid; when the value of P EV is negative, it means that the electric vehicle provides electric energy to the power grid in the reverse direction.

[0117] Electric vehicle charge and discharge constraint:

[0118]

[0119]

[0120]

[0121]

[0122]

[0123]

[0124]

[0125] Among them, is the state of charge of the vth electric vehicle at time t; are the upper and lower limits of the state of charge of the vth electric vehicle respectively; is the charging power of the vth electric vehicle at time t; are the charging and discharging power limits of the vth electric vehicle respectively; respectively represent whether the vth electric vehicle is in the charging or discharging state at time t. If it is, set it to 1, otherwise set it to 0; represents whether the vth electric vehicle is connected to the power grid at time t; are the initial and final states of charge of the vth electric vehicle respectively; is the charging efficiency of the vth electric vehicle.

[0126] Based on the above simultaneous calculation of the minimum grid cost and the battery charge and discharge model, the final charge and discharge plan is obtained.

[0127] The present application also provides an electric vehicle charging scheduling device, including:

[0128] A grid module, configured to define the area to be scheduled as a scheduling space and divide it into multiple grids according to longitude and latitude;

[0129] An acquisition module, configured to acquire the driving trajectory data composed of the longitude, latitude and time of multiple vehicles within the scheduling space;

[0130] A matrix module, configured to construct a grid heat matrix according to the driving trajectory data and the grids;

[0131] A decomposition module, configured to perform singular value decomposition on the grid heat matrix to generate a semantic space composed of several left singular orthogonal vectors, and the semantic space implies the degree of interest of drivers in each grid;

[0132] A clustering module, configured to cluster the driving trajectory data according to the semantic space to generate driving trajectory cluster data;

[0133] A model module, configured to establish a battery charge and discharge model for electric vehicles with the goal of minimizing the difference between the charging cost and the discharge benefit;

[0134] A screening module, configured to use the Monte Carlo method to screen the variable parameters in the battery charge and discharge model with the driving trajectory cluster data as the initial variable parameter combination to determine the optimal variable parameter combination;

[0135] A calculation module, configured to calculate an electric vehicle charging scheduling scheme based on the battery charge and discharge model with the optimal variable parameter combination.

Claims

1. An electric vehicle charging scheduling method, characterized in that, Including: Define the area to be scheduled as the scheduling space and divide it into multiple grids according to longitude and latitude; Collect the driving trajectory data composed of the longitude, latitude and time of multiple vehicles in the scheduling space; Construct a grid heat matrix based on the driving trajectory data and the grids; Perform singular value decomposition on the grid heat matrix to generate a semantic space composed of several left singular orthogonal vectors, and the semantic space implies the degree of interest of drivers in each grid; Cluster the driving trajectory data according to the semantic space to generate driving trajectory cluster data; Establish a battery charge and discharge model for electric vehicles with the goal of minimizing the difference between charging cost and discharge income; Use the Monte Carlo method to screen the variable parameters in the battery charge and discharge model with the driving trajectory cluster data as the initial variable parameter combination to determine the optimal variable parameter combination; Based on the battery charge and discharge model with the optimal variable parameter combination, calculate the charging scheduling plan for electric vehicles.

2. The electric vehicle charging scheduling method according to claim 1, wherein Constructing a grid heat matrix based on the driving trajectory data and the grids includes: Construct a grid-vehicle matrix based on the driving trajectory data and the grids, and each element in the grid-vehicle matrix represents the residence time of the vehicle in the corresponding grid; Record the elements with residence time less than the preset time length as 0, and convert the grid-vehicle matrix into a grid heat matrix.

3. The electric vehicle charging scheduling method according to claim 1, wherein The degree of interest is represented by the inner product of the eigenvector after singular value decomposition and the corresponding singular value.

4. The electric vehicle charging scheduling method according to claim 1, wherein, The battery charge and discharge model includes: Among them, is the charging cost, is the discharging income, is the battery loss cost, and are respectively the charging cost and discharging income of the electric vehicle after reaching the destination at the k-th trip, and are binary variables, and when their values are 1, they respectively indicate that the electric vehicle charges or discharges at time t after reaching the destination at the k-th trip, and are respectively the charging electricity price and discharging electricity price at time t, is the charging and discharging power of the electric vehicle at the destination of the k-th trip C b is the battery replacement price, N b is the maximum number of charge and discharge cycles of the battery, N ch and N di are respectively the number of charging times and discharging times of the battery in a day, η ch and η di are respectively the charging efficiency and discharging efficiency.

5. The electric vehicle charging scheduling method according to claim 1, characterized in that, Using the Monte Carlo method to screen the variable parameters in the battery charge and discharge model, the determination steps of the number of iterations include: Generate fitness according to the initial variable parameter combination; According to the fitness, perform evolutionary calculation using an evolutionary algorithm, including: defining the number of invalid evolutions and initializing it to 0; after each evolution of the evolutionary algorithm is completed, a new fitness is generated. When the absolute value of the difference between the nth fitness and the (n-1)th fitness is less than the preset evolution accuracy, the number of invalid evolutions is incremented by 1; when the absolute value of the difference between the nth fitness and the (n-1)th fitness is greater than or equal to the preset evolution accuracy, the number of invalid evolutions is cleared and the evolutionary calculation is stopped.

6. The electric vehicle charging scheduling method according to claim 1, characterized in that, Using the Monte Carlo method to screen the variable parameters in the scheduling model includes: Model the uncertainty of the variable parameter combination of the battery charge and discharge model, and use Bayes' formula to select the variable parameter combination that is most consistent with the observed value to update the uncertainty of the next iteration; When the iteration is completed, obtain the variable parameter combination after iteration to complete the screening.

7. The electric vehicle charging scheduling method according to claim 1, characterized in that, Performing singular value decomposition on the grid heat matrix to generate a semantic space composed of several left singular orthogonal vectors includes: According to the singular value decomposition formula E = M·diag(λ1, λ2,..., λ s )·N', perform singular value decomposition on the grid heat matrix; where λ i is the singular value of matrix E; the eigenvectors of EE’ are the column vectors of M; the eigenvectors of E’E are the column vectors of N; s is the rank of matrix E; After obtaining the column vectors of M, use the left singular vectors, that is, the column vectors of M, to generate the semantic space, including: for each left singular vector M_i, calculate its inner product with the corresponding singular value σ_i.

8. An electric vehicle charging scheduling device, characterized in that, Including: A grid module for defining the area to be scheduled as the scheduling space and dividing it into multiple grids according to longitude and latitude; A collection module, configured to collect driving trajectory data composed of the longitude, latitude and time of multiple vehicles within the scheduling space; A matrix module, configured to construct a grid heat matrix according to the driving trajectory data and the grid; A decomposition module, configured to perform singular value decomposition on the grid heat matrix to generate a semantic space composed of a plurality of left singular orthogonal vectors, where the semantic space implies the degree of interest of drivers in each grid; A clustering module, configured to cluster the driving trajectory data according to the semantic space to generate driving trajectory cluster data; A model module, configured to establish a battery charge and discharge model of an electric vehicle with the goal of minimizing the difference between the charging cost and the discharge benefit; A screening module, configured to use the Monte Carlo method to screen the variable parameters in the battery charge and discharge model with the driving trajectory cluster data as the initial variable parameter combination to determine the optimal variable parameter combination; A calculation module, configured to calculate an electric vehicle charging scheduling scheme based on the battery charge and discharge model with the optimal variable parameter combination.

9. An electric vehicle charging scheduling device, characterized in that, Including: A memory, configured to store a computer executable program for the electric vehicle charging scheduling method according to any one of claims 1 to 7; A processor, configured to retrieve the computer executable program from the memory and execute: defining the area to be scheduled as a scheduling space and dividing it into multiple grids according to longitude and latitude; collecting driving trajectory data composed of the longitude, latitude and time of multiple vehicles within the scheduling space; Constructing a grid heat matrix according to the driving trajectory data and the grid; Performing singular value decomposition on the grid heat matrix to generate a semantic space composed of a plurality of left singular orthogonal vectors, where the semantic space implies the degree of interest of drivers in each grid; clustering the driving trajectory data according to the semantic space to generate driving trajectory cluster data; establishing a battery charge and discharge model of an electric vehicle with the goal of minimizing the difference between the charging cost and the discharge benefit; using the Monte Carlo method to screen the variable parameters in the battery charge and discharge model with the driving trajectory cluster data as the initial variable parameter combination to determine the optimal variable parameter combination; calculating an electric vehicle charging scheduling scheme based on the battery charge and discharge model with the optimal variable parameter combination.

10. A storage medium, characterized in that, Stored with a computer executable program, which is used to be retrieved by a processor and execute the steps of the electric vehicle charging scheduling method according to any one of claims 1 to 7.

Images