Vehicle charging demand migration control method based on multi-source spatiotemporal data fusion

Abstract

The present application relates to a vehicle charging demand migration control method based on multi-source spatiotemporal data fusion. The method includes: obtaining real-time status data of the area to be predicted at the target time, as well as multi-source data in a first historical time period and demand forecast errors in a second historical time period; performing spatiotemporal alignment processing on the multi-source data according to preset spatiotemporal units to obtain a multi-dimensional spatiotemporal feature matrix; the multi-dimensional spatiotemporal feature matrix is ​​used to represent the multi-source data of multiple sub-areas of the area to be predicted in multiple sub-historical time periods in the first historical time period; obtaining the target prediction value of each sub-area based on the real-time status data, demand forecast errors, the multi-dimensional spatiotemporal feature matrix and the trained spatiotemporal coupling prediction model; the target prediction value is the electric energy demand for vehicle charging in the sub-area during the time period to be predicted; and determining the migration control strategy of the area to be predicted during the time period to be predicted based on each target prediction value. The use of this method can improve the accuracy of prediction.

Description

Technical Field

[0001] The present application relates to the technical field of charging demand prediction, and in particular to a vehicle charging demand migration control method based on multi-source spatiotemporal data fusion. Background Art

[0002] New energy vehicles have the advantages of high energy efficiency, low operating costs and easy maintenance, which have enabled them to gradually replace fuel vehicles and become the mainstream means of transportation in modern society.

[0003] However, the contradiction between the charging demand of new energy vehicles and the number of charging stations is becoming increasingly prominent. Accurately predicting the charging demand of electric vehicles is the most basic management method in the management of new energy vehicle charging networks, but the prediction accuracy of new energy vehicle charging demand prediction methods in related technologies is relatively low. Summary of the Invention

[0004] Based on this, it is necessary to provide a vehicle charging demand migration control method based on multi-source spatiotemporal data fusion that can improve prediction accuracy in response to the above technical problems.

[0005] In the first aspect, the present application proposes a vehicle charging demand migration control method based on multi-source spatiotemporal data fusion, the method comprising:

[0006] According to the forecast time period of the forecast area, real-time status data of the forecast area at the target time is obtained, as well as multi-source data in the first historical time period and demand forecast errors in the second historical time period;

[0007] Performing spatiotemporal alignment processing on the multi-source data according to preset spatiotemporal units to obtain a multi-dimensional spatiotemporal feature matrix; the multi-dimensional spatiotemporal feature matrix is ​​used to represent the multi-source data of multiple sub-regions of the area to be predicted in multiple sub-historical time periods in the first historical time period;

[0008] Obtaining a target prediction value for each of the sub-areas based on the real-time status data, the demand prediction error, the multi-dimensional spatiotemporal feature matrix, and the trained spatiotemporal coupling prediction model; the target prediction value is the electric energy demand for vehicle charging in the sub-area during the time period to be predicted;

[0009] According to each of the target prediction values, a migration control strategy for the area to be predicted in the time period to be predicted is determined.

[0010] In one embodiment, obtaining the target prediction value of each sub-region according to the real-time status data, the demand prediction error, the multi-dimensional spatiotemporal feature matrix, and the trained spatiotemporal coupling prediction model includes:

[0011] Acquiring historical charging transaction data of the area to be predicted within a third historical time period according to the area to be predicted;

[0012] Obtaining an initial prediction value for each of the sub-regions according to the multi-dimensional spatiotemporal feature matrix and the spatiotemporal coupling prediction model;

[0013] Obtaining a revised predicted value for each of the sub-areas based on the real-time status data, the historical charging transaction data, and the initial predicted value for each of the sub-areas;

[0014] A target forecast value for each of the sub-regions is obtained according to each of the corrected forecast values, the real-time status data, and the demand forecast error.

[0015] In one embodiment, the historical charging transaction data includes the number of charging migrations between multiple sub-areas of the area to be predicted; and obtaining the revised predicted value of each sub-area based on the real-time status data, the historical charging transaction data, and the initial predicted value of each sub-area includes:

[0016] Determine the initial migration probability between each sub-area according to the number of charge migrations between each sub-area;

[0017] According to the real-time status data, the initial migration probabilities are modified to obtain target migration probabilities;

[0018] According to the target migration probability and each of the initial prediction values, a revised prediction value of each of the sub-regions is obtained.

[0019] In one embodiment, the real-time status data includes the real-time electricity price, the total number of charging piles, and the number of idle charging piles in each sub-area; and the modifying of each initial migration probability based on the real-time status data to obtain the target migration probability includes:

[0020] Based on the real-time electricity prices of each sub-region, obtain the electricity price difference data between sub-regions;

[0021] The target migration probability between each sub-area is obtained based on the preset first sensitivity coefficient, the second sensitivity coefficient, the initial migration probabilities, the total number of charging piles and the number of idle charging piles in each sub-area, and the electricity price difference data between each sub-area.

[0022] In one embodiment, obtaining the target forecast value of each sub-region according to each of the corrected forecast values, the real-time status data, and the demand forecast error includes:

[0023] Optimizing the spatiotemporal coupling prediction model according to the real-time status data and the demand forecast error;

[0024] According to each of the corrected prediction values ​​and the optimized spatiotemporal coupling prediction model, a target prediction value for each of the sub-regions is obtained.

[0025] In one embodiment, the real-time status data includes a real-time electricity price, a demand overload index, and a traffic congestion index of each sub-region; and optimizing the spatiotemporal coupling prediction model based on the real-time status data and the demand forecast error includes:

[0026] Get the average electricity price based on the real-time electricity price of each sub-region;

[0027] Determining real-time electricity price fluctuation data for each sub-region based on the real-time electricity price of each sub-region and the average electricity price;

[0028] Constructing a reward function for each sub-region based on the preset first weight coefficient, second weight coefficient, and third weight coefficient, as well as the demand forecast error, real-time electricity price fluctuation data, and demand overload index of each sub-region;

[0029] The spatiotemporal coupling prediction model is optimized according to the reward function of each sub-area, the demand forecast error, the traffic congestion index, the demand overload index and the real-time electricity price fluctuation data.

[0030] In one embodiment, determining the migration control strategy of the to-be-predicted area in the to-be-predicted time period according to each of the target prediction values ​​includes:

[0031] Obtaining the power reserve data of each sub-area in the area to be predicted;

[0032] When the ratio of the target prediction value of the sub-region to the corresponding electric energy reserve data is less than a first threshold, determining the sub-region as a first sub-region and reducing the real-time electricity price of the first sub-region;

[0033] If the ratio of the target prediction value of the sub-region to the corresponding electric energy reserve data is greater than or equal to the first threshold and less than the second threshold, determine the sub-region as the second sub-region and keep the real-time electricity price of the second sub-region unchanged;

[0034] When the ratio of the target prediction value of the sub-area to the corresponding electric energy reserve data is greater than or equal to a second threshold value, the sub-area is determined to be a third sub-area, and the real-time electricity price of the sub-area is updated according to the target prediction value of the third sub-area and the corresponding electric energy reserve data, and / or energy is dispatched from the first sub-area to the third sub-area according to the target prediction value of the third sub-area and the corresponding electric energy reserve data, as well as the idle electric energy data of the first sub-area; wherein the idle electric energy data is related to the target prediction value of the first sub-area and the corresponding electric energy reserve data.

[0035] In a second aspect, the present application proposes a vehicle charging demand migration control device based on multi-source spatiotemporal data fusion, the device comprising:

[0036] A first execution module is configured to obtain, based on a to-be-predicted time period of the to-be-predicted area, real-time status data of the to-be-predicted area at a target time, as well as multi-source data within a first historical time period and a demand forecast error within a second historical time period;

[0037] a second execution module, configured to perform spatiotemporal alignment processing on the multi-source data according to a preset spatiotemporal unit to obtain a multi-dimensional spatiotemporal feature matrix; the multi-dimensional spatiotemporal feature matrix is ​​used to represent the multi-source data of multiple sub-regions of the area to be predicted in multiple sub-historical time periods in the first historical time period;

[0038] a third execution module, configured to obtain a target prediction value for each of the sub-areas based on the real-time status data, the demand prediction error, the multi-dimensional spatiotemporal feature matrix, and the trained spatiotemporal coupling prediction model, the target prediction value being the electric energy demand for vehicle charging in the sub-area during the time period to be predicted;

[0039] The fourth execution module is used to determine the migration control strategy of the area to be predicted in the time period to be predicted according to each of the target prediction values.

[0040] In a third aspect, the present application proposes a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method described in any one of the above embodiments when executing the computer program.

[0041] In a fourth aspect, the present application proposes a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method described in any one of the above embodiments when the computer program is executed by a processor.

[0042] The above-mentioned vehicle charging demand migration control method, device, computer equipment and computer-readable storage medium based on multi-source spatiotemporal data fusion first obtains the real-time status data of the area to be predicted at the target moment according to the time period to be predicted of the area to be predicted, as well as the multi-source data of the area to be predicted in the first historical time period and the demand prediction error in the second historical time period. Afterwards, the multi-source data is subjected to spatiotemporal alignment processing according to the preset spatiotemporal units to obtain a multi-dimensional spatiotemporal feature matrix; then, based on the real-time status data, the demand prediction error, the multi-dimensional spatiotemporal feature matrix and the trained spatiotemporal coupling prediction model, the target prediction value of each sub-area is obtained. Finally, based on each target prediction value, the migration control strategy of the area to be predicted in the time period to be predicted is determined. The present application dynamically predicts the charging demand for the time period to be predicted by combining the multi-source data and prediction errors of multiple historical time periods, which can improve the prediction accuracy and the prediction precision. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1 1 is a flow chart of a method for regulating and controlling vehicle charging demand migration based on multi-source spatiotemporal data fusion in one embodiment;

[0045] Figure 2 Schematic diagram of the process of step S103 in one embodiment;

[0046] Figure 3 Schematic diagram of the process of step S203 in one embodiment;

[0047] Figure 4 Schematic diagram of the process of step S302 in one embodiment;

[0048] Figure 5 Schematic diagram of the process of step S204 in one embodiment;

[0049] Figure 6 Schematic diagram of the process of step S501 in one embodiment;

[0050] Figure 7 Schematic diagram of the process of step S104 in one embodiment;

[0051] Figure 8 This is a workflow diagram of a vehicle charging demand migration control method based on multi-source spatiotemporal data fusion in one embodiment;

[0052] Figure 9 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0054] As mentioned in the background technology section, with the tightening of environmental regulations and the advancement of new energy technologies, the new energy vehicle market has seen explosive growth. The spatiotemporal distribution characteristics of its charging demand have had a profound impact on power system operation, charging infrastructure planning, and user experience. However, there is a structural contradiction between the current layout of charging facilities and the actual needs of new energy vehicle users. Traditional static planning methods are difficult to adapt to the dynamic changes in charging demand, resulting in low charging pile utilization and local power supply shortages in some areas. In addition, the randomness of user travel behavior, the diversity of battery technologies, and the complexity of grid load fluctuations further exacerbate the dynamic uncertainty of charging demand, and there is an urgent need to achieve precise scheduling of charging resources through modeling and forecasting technologies.

[0055] At the same time, breakthroughs in digital technology are providing new opportunities to address this challenge. The widespread adoption of the Internet of Things (IoT) and the Internet of Vehicles (IoV) allows for the real-time collection of data such as vehicle driving trajectories and battery status. Combining big data analysis with artificial intelligence algorithms can deeply explore the temporal and spatial patterns of charging demand. By building a dynamic modeling and forecasting system, not only can the location and configuration of charging stations be optimized, improving the convenience of charging services, but it can also assist grid operators in developing flexible scheduling strategies to mitigate the impact of large-scale charging on the power grid. This, in turn, promotes the sustainable development of the electric vehicle industry while ensuring the efficient and stable operation of the energy system.

[0056] Traditional charging demand forecasting (such as a method and system for predicting urban electric vehicle charging demand based on heterogeneous multi-graph convolutional network fusion, CN119273046A) relies solely on historical electricity consumption data and some user behavior data (such as driving trajectories and charging time preferences), without incorporating dynamic influencing factors such as traffic conditions (congestion index) and meteorological conditions (temperature, rainfall). This results in a large deviation between the prediction results and actual demand. Existing models (such as a method and device for predicting electric vehicle charging demand based on urban areas, CN114707747A) extract spatiotemporal characteristics of the region but only consider geographic distance. They do not quantify the impact of real-time factors such as regional electricity price differences and charging pile capacity constraints on cross-regional demand migration. A current patent (a system and method for electric vehicle charging scheduling based on deep reinforcement learning and renewable energy, CN119417129A) uses reinforcement learning to optimize charging scheduling strategies, but the multi-source heterogeneous data is not effectively integrated, the data value is insufficiently mined, and the coordinated optimization of prediction errors and grid risks is not achieved.

[0057] To solve the above technical problems, in an exemplary embodiment, please refer to Figure 1 , this application proposes a vehicle charging demand migration control method based on multi-source spatiotemporal data fusion, the method includes steps S101 to S104.

[0058] S101: According to a to-be-predicted time period of the to-be-predicted area, real-time status data of the to-be-predicted area at a target time, as well as multi-source data in a first historical time period and a demand forecast error in a second historical time period are obtained.

[0059] Among them, the area to be predicted includes multiple sub-areas, the time period to be predicted includes at least one sub-time period, the time step of each sub-time period is the same, the target moment is the starting moment of the time period to be predicted, that is, the target moment is the starting moment of the first sub-time period in the time period to be predicted, the time interval of the time period to be predicted falls within the time interval of the first historical time period, and the time period to be predicted is after the first historical time period, and the second historical time period is the previous time part of the time period to be predicted, that is, the target moment is the end moment of the second historical time period. In an example, the time step is fifteen minutes, and the time period to be predicted may be from 10:00 to 11:00 a.m. on April 15th. The time period to be predicted includes four sub-time periods, namely, from 10:00 to 10:15 a.m. on April 15th, from 10:15 to 10:30 a.m. on April 15th, from 10:30 to 10:45 a.m. on April 15th, and from 10:45 to 11:00 a.m. on April 15th; the first historical time period is from 10:00 to 11:00 a.m. on April 14th, and the first historical time period also includes four sub-time periods, namely, from 10:00 to 10:15 a.m. on April 14th, from 10:15 to 10:30 a.m. on April 14th, from 10:30 to 10:45 a.m. on April 14th, and from 10:45 to 11:00 a.m. on April 14th; the second historical time period is from 9:45 to 10:00 a.m. on April 15th.

[0060] The multi-source data for the first historical time period refers to the multi-source data for each sub-region of the predicted area within each sub-time period of the first historical time period. The multi-source data may include at least user behavior data, power grid operation data, traffic situation data, meteorological data, and charging facility data. User behavior data specifically includes charging timestamps, driving trajectories, and charging amounts. Power grid operation data specifically includes actual load and time-of-use electricity prices. Traffic situation data specifically includes traffic congestion index. Meteorological data specifically includes temperature and rainfall (specifically, the impact of temperature and rainfall on charging demand). Charging facility data specifically includes charging pile usage status. The demand forecast error for the second historical time period refers to the difference between the final predicted value of the electric energy demand for the second historical time period and the actual electric energy demand for the second historical time period.

[0061] S102: Performing spatiotemporal alignment processing on the multi-source data according to preset spatiotemporal units to obtain a multi-dimensional spatiotemporal feature matrix; the multi-dimensional spatiotemporal feature matrix is ​​used to represent the multi-source data of multiple sub-regions of the area to be predicted in multiple sub-historical time periods in the first historical time period.

[0062] Among them, the spatiotemporal unit includes a geographic grid and a time step. The area to be predicted is divided into multiple sub-areas according to the geographic grid. After obtaining the multi-source data of the first historical time period, the multi-source data of the first historical time period can be aligned according to the preset spatiotemporal unit to form a multidimensional spatiotemporal feature matrix. ,in is the number of sub-regions, is the time step, For example, the area to be predicted is divided into four sub-areas according to the geographic grid. The time section is fifteen minutes. The multi-source data includes nine feature dimensions: charging timestamp, driving trajectory, charging amount, actual load, time-of-use electricity price, traffic congestion index, temperature, rainfall and charging pile group usage status. The first historical time period includes four sub-time periods, thus forming a multi-dimensional spatiotemporal feature matrix. .

[0063] S103: Obtain a target prediction value for each sub-area based on real-time status data, demand prediction error, multi-dimensional spatiotemporal feature matrix, and the trained spatiotemporal coupling prediction model; the target prediction value is the electric energy demand for vehicle charging in the sub-area during the predicted time period.

[0064] Furthermore, the prediction value of each sub-region can be obtained based on the dimensional space-time feature matrix and the trained space-time coupling prediction model. At the same time, the space-time coupling prediction model can be optimized according to the real-time status data and demand prediction error. Then, the target prediction value of each sub-region can be determined based on the prediction value and the optimized space-time coupling prediction model.

[0065] S104: Determine a migration control strategy for the area to be predicted during the time period to be predicted based on each target prediction value.

[0066] After obtaining the target prediction value for each sub-region, a migration control strategy can be specified for each sub-region. For example, the electricity price in the sub-region with higher charging demand can be raised, and the electricity price in the sub-region with lower charging demand can be lowered to guide vehicles to charge in the sub-region with lower charging demand.

[0067] The above-mentioned vehicle charging demand migration control method based on multi-source spatiotemporal data fusion first obtains the real-time status data of the to-be-predicted area at the target moment according to the to-be-predicted time period of the to-be-predicted area, as well as the multi-source data of the to-be-predicted area in the first historical time period and the demand prediction error in the second historical time period. Afterwards, the multi-source data is subjected to spatiotemporal alignment processing according to the preset spatiotemporal units to obtain a multi-dimensional spatiotemporal feature matrix; then, the target prediction value of each of the sub-areas is obtained according to the real-time status data, the demand prediction error, the multi-dimensional spatiotemporal feature matrix and the trained spatiotemporal coupling prediction model. Finally, the migration control strategy of the to-be-predicted area in the to-be-predicted time period is determined according to each of the target prediction values. The present application dynamically predicts the charging demand for the to-be-predicted time period by combining the multi-source data and prediction errors of multiple historical time periods, which can improve the prediction accuracy and improve the prediction precision.

[0068] In an exemplary embodiment, see Figure 2, step S103, obtains the target prediction value of each sub-region according to the real-time status data, demand forecast error, multi-dimensional spatiotemporal feature matrix and the trained spatiotemporal coupling prediction model, including steps S201 to S204.

[0069] S201: Acquire historical charging transaction data of the area to be predicted within a third historical time period according to the area to be predicted time period.

[0070] To improve the accuracy of the prediction results, the application can also obtain historical charging transaction data for the predicted area within a third historical time period, where the time interval of the first historical time period falls within the time interval of the third historical time period. For example, if the predicted time period is from 10:00 to 11:00 on April 15th, the first historical time period is from 10:00 to 11:00 on April 14th, and the third historical time period can be from 12:00 on April 8th to 23:59 on April 14th.

[0071] S202: Obtaining initial prediction values ​​for each sub-region based on the multi-dimensional spatiotemporal feature matrix and the spatiotemporal coupling prediction model.

[0072] Before step S202, it is also necessary to pre-build a spatiotemporal coupling prediction model. In one example, the spatiotemporal coupling prediction model can be a spatiotemporal graph convolutional network (ST-GCN, Spatial Temporal Graph Convolutional Networks) model. The spatiotemporal graph convolutional network model includes a spatial graph convolution layer, a temporal convolution layer and an LSTM (Long Short-term Memory Networks, Long Short-term Memory Neural Network) module. Among them, when building the spatial graph convolution layer, it is necessary to build a regional adjacency matrix based on the geographical distance of each sub-region in the area to be predicted. , for example, the calculation formula of the regional adjacency matrix between sub-region i and sub-region j is ,in is the adjustable attenuation coefficient. Through the gradient descent algorithm in the training process, adjustments are made to adapt to the spatial correlation characteristics of different sub-regions and optimize the model's ability to model spatial dependencies. is the Euclidean distance between sub-region i and sub-region j (given by longitude and latitude coordinates Calculated, are the longitude and latitude coordinates of sub-region i, is the longitude and latitude coordinates of sub-region j), the regional adjacency matrix is ​​used to characterize the spatial correlation between sub-regions, quantify the mutual influence of charging demands in different sub-regions by weighting them with geographical distance, capture the spatial dependency of charging demands between sub-regions, and provide a basis for the model to learn the spatial correlation between sub-regions. The temporal convolution layer uses a one-dimensional convolution kernel to extract time series dependencies, and the number of output channels can be 64. The LSTM module is used to capture the periodic characteristics of user charging behavior, and the hidden layer dimension can be 128. By inputting the multi-dimensional spatiotemporal feature matrix into the spatiotemporal coupling prediction model, the initial prediction value of each sub-region in the prediction time period can be obtained. .

[0073] S203: Obtain a revised prediction value for each sub-region based on the real-time status data, the historical charging transaction data, and the initial prediction value for each sub-region.

[0074] In the application, after obtaining the initial prediction value of each sub-area, the migration probability between each sub-area can be calculated based on the real-time status data and historical charging transaction data. The initial prediction value of each sub-area can be corrected according to the migration probability between each sub-area to obtain the corrected prediction value of each sub-area.

[0075] S204: Obtain a target forecast value for each sub-region based on each corrected forecast value, real-time status data, and demand forecast error.

[0076] While calculating the revised prediction value of each sub-region, the spatiotemporal coupling prediction model can be optimized according to the real-time status data and demand prediction error, and then the target prediction value of each sub-region can be determined based on the optimized spatiotemporal coupling prediction model and the revised prediction value of each sub-region.

[0077] In an exemplary embodiment, the historical charging transaction data includes the number of charging migrations between multiple sub-areas of the area to be predicted. Figure 3 , step S203, obtaining the revised prediction value of each sub-area according to the real-time status data, the historical charging transaction data and the initial prediction value of each sub-area, including steps S301 to S303.

[0078] S301: Determine the initial migration probability between each sub-area according to the number of charge migrations between each sub-area.

[0079] The historical charging transaction data may include the IDs of all vehicles that charged within the predicted area during the third historical time period, the departure area, charging area, and charging timestamp of each vehicle's each charging event. The number of charging migrations between sub-areas for all vehicles that charged within the predicted area during the third historical time period is then determined based on the departure area, charging area, and charging timestamp of each vehicle's each charging event. The initial migration probability between sub-areas is then calculated. For example, the initial migration probability between sub-areas can be calculated using the following formula: ,in, is the initial migration probability from sub-region i to sub-region j, is the number of charging migrations of vehicles that migrate from sub-area i to sub-area j among all vehicles that have been charged in the third historical time period in the area to be predicted, The sum of the number of charging migrations of all vehicles in the predicted area that have been charged during the third historical time period and that have migrated from sub-area i to other sub-areas in the predicted area except sub-area i. Migration refers to the behavior of moving between areas before charging.

[0080] S302: Correct each initial transition probability according to the real-time status data to obtain a target transition probability.

[0081] Afterwards, the initial migration probability can be modified by incorporating real-time grid capacity constraints and electricity price differences. Specifically, real-time status data includes information such as the real-time electricity price, total number of charging piles, and number of idle charging piles in each sub-region. Based on the real-time status of each sub-region, the initial migration probability is modified to obtain the target migration probability.

[0082] S303: Obtain a revised prediction value for each sub-region according to the target migration probability and each initial prediction value.

[0083] Afterwards, the demand distribution between sub-regions can be adjusted according to the target migration probability, and the cross-region demand flow can be calculated. ,in, is the demand flow from sub-region i to sub-region j, is the initial migration probability from sub-region i to sub-region j, is the target migration probability from sub-region i to sub-region j, and then the modified prediction value of each sub-region can be obtained, for example, the modified prediction value from sub-region i to sub-region j The calculation formula is: .

[0084] In an exemplary embodiment, the real-time status data includes the real-time electricity price, the total number of charging piles, and the number of idle charging piles in each sub-area. Figure 4, step S302, correcting each initial migration probability according to the real-time status data to obtain the target migration probability, including step S401 and step S402.

[0085] S401: Obtain electricity price difference data between sub-regions based on the real-time electricity price of each sub-region.

[0086] For example, the real-time electricity price of sub-region i is , the real-time electricity price of sub-region j is , then the electricity price difference data between sub-area i and sub-area j can be calculated .

[0087] S402: Obtain target migration probabilities between sub-areas based on a preset first sensitivity coefficient, a second sensitivity coefficient, each initial migration probability, the total number of charging piles and the number of idle charging piles in each sub-area, and the electricity price difference data between sub-areas.

[0088] Then, the target migration probability from sub-region i to sub-region j can be calculated according to the following formula: ,in, is the first sensitivity coefficient, is the second sensitivity coefficient, is the initial migration probability from sub-region i to sub-region j, is the number of available charging piles in area j, is the total number of charging piles in area j, is the electricity price difference data between sub-area i and sub-area j.

[0089] In application, the first sensitivity coefficient and the second sensitivity coefficient , can be determined by constructing the loss function , where KL divergence measures the difference between the historical and revised probability distributions; gradient descent method is used for iterative optimization and , until the loss function converges. The specific execution steps are as follows: ① Initialize the first sensitive coefficient and the second sensitivity coefficient 1; ② Using the current sensitivity coefficient and Calculate target migration probability ; ③Calculate the initial migration probability And calculate the KL divergence loss of the two ;④Calculate the loss function about 、 Gradient 、 ; ⑤ Update the sensitivity coefficient using gradient descent method and : , ,in is the learning rate, set to 0.001; ⑥ Check whether the loss function converges, if ( is the preset convergence threshold, here it is 10 -5 ), then stop the iteration, otherwise return to step ② to continue execution.

[0090] In an exemplary embodiment, see Figure 5 , step S204, obtaining the target prediction value of each sub-region according to each corrected prediction value, real-time status data and demand prediction error, including step S501 and step S502.

[0091] S501: Optimize the spatiotemporal coupling prediction model based on real-time status data and demand prediction error.

[0092] It is understandable that there may be a certain error between the predicted value and the actual value of the charging demand. Therefore, in this application, the demand prediction error of the previous time step of the time period to be predicted and the real-time status data at the starting moment of the time period to be predicted will be used to optimize the time-space coupling prediction model to improve the accuracy of the prediction of the time-space coupling prediction model.

[0093] S502: Obtain a target prediction value for each sub-region according to each corrected prediction value and the optimized spatiotemporal coupling prediction model.

[0094] Specifically, each revised forecast value is used as the original data input of the spatiotemporal coupling forecast model before optimization, and then the parameters of the spatiotemporal coupling forecast model are dynamically optimized according to the real-time status data and demand forecast error, so that the result approaches the direction of higher reward value score, and the optimized target forecast value is obtained. .

[0095] In an exemplary embodiment, the real-time status data includes the real-time electricity price, demand overload index and traffic congestion index of each sub-region; see Figure 6 , step S501, optimizing the spatiotemporal coupling prediction model according to the real-time status data and the demand prediction error, including steps S601 to S604.

[0096] S601: Obtain an average electricity price based on the real-time electricity price of each sub-region.

[0097] First, the average electricity price of the area to be predicted is obtained based on the real-time electricity price of each sub-area .

[0098] S602: Determine the real-time electricity price fluctuation data of each sub-region based on the real-time electricity price and the average electricity price of each sub-region.

[0099] Then, according to the average electricity price of the area to be predicted And the real-time electricity price of each sub-region, calculate the real-time electricity price fluctuation data of each sub-region. For example, the real-time electricity price of a sub-region is , then the real-time electricity price fluctuation data of the sub-region is .

[0100] S603: Constructing a reward function for each sub-region based on the preset first weight coefficient, second weight coefficient, and third weight coefficient, as well as the demand forecast error, real-time electricity price fluctuation data, and demand overload index of each sub-region.

[0101] After that, define the state space of each sub-region and construct the reward function of each sub-region. Taking a sub-region as an example, the state space of the sub-region is , the reward function of this sub-region ,in, is the demand forecast error for this sub-region, is the demand overload index ( = (time) indicates that the actual power demand of the sub-region exceeds 90% of the actual power reserve data of the sub-region at the target time. is the traffic congestion index; the reward function of the sub-area is , 、 and are the first weight coefficient, the second weight coefficient and the third weight coefficient respectively. In one example, =0.8, =0.3, =0.2, the optimization objective of the reward function is to minimize the prediction error and grid risk.

[0102] S604: Optimize the spatiotemporal coupling prediction model based on the reward function, demand forecast error, traffic congestion index, demand overload index, and real-time electricity price fluctuation data of each sub-region.

[0103] Among them, the Double DQN algorithm is used to update the model parameters. The action space is the fine-tuning step size (±0.001) of the ST-GCN convolution kernel weight and the LSTM forget gate parameter. The corrected prediction values ​​are used as the original data input of the spatiotemporal coupling prediction model before optimization. Then, the parameters of the spatiotemporal coupling prediction model are dynamically optimized according to the reward function and state space, so that the result approaches the direction of higher reward value scores, and the optimized target prediction value is obtained. .

[0104] The reinforcement learning training process for the spatiotemporal coupling prediction model includes: initializing the experience replay pool with a capacity of 10,000 state-action-reward records; synchronizing the target network parameters every 1,000 steps to prevent overestimation of the Q value; setting the exploration rate Decays from 0.5 to 0.1 over time, balancing exploration and exploitation.

[0105] In an exemplary embodiment, see Figure 7 , step S104, determining the migration control strategy of the area to be predicted in the time period to be predicted according to each target prediction value, including steps S701 to S704.

[0106] S701: Obtaining power reserve data of each sub-area in the area to be predicted.

[0107] The electric energy reserve data of the sub-region refers to the electric energy actually available and reserved in each sub-region at the target time.

[0108] S702: When the ratio of the target prediction value of the sub-region to the corresponding electric energy reserve data is less than a first threshold, determine the sub-region as a first sub-region, and reduce the real-time electricity price of the first sub-region.

[0109] For subregions where the ratio of the target predicted value to the corresponding electric energy reserve data is less than a first threshold, they can be marked as first subregions, where the first threshold can be 0.5. For the first subregion, the real-time electricity price for the first subregion at the target time can be lowered by a preset ratio, which can be 80%.

[0110] S703: When the ratio of the target prediction value of the sub-region to the corresponding electric energy reserve data is greater than or equal to the first threshold and less than the second threshold, determine the sub-region as the second sub-region and keep the real-time electricity price of the second sub-region unchanged.

[0111] For a sub-region where the ratio of the target prediction value to the corresponding electric energy reserve data is greater than or equal to the first threshold and less than a preset second threshold, it can be marked as a second sub-region, and the second threshold can be 0.8.

[0112] S704: When the ratio of the target prediction value of the sub-region to the corresponding electric energy reserve data is greater than or equal to the second threshold, the sub-region is determined to be the third sub-region, and the real-time electricity price of the sub-region is updated according to the target prediction value of the third sub-region and the corresponding electric energy reserve data, and / or energy is dispatched from the first sub-region to the third sub-region according to the target prediction value of the third sub-region and the corresponding electric energy reserve data, as well as the idle electric energy data of the first sub-region; wherein the idle electric energy data is related to the target prediction value of the first sub-region and the corresponding electric energy reserve data.

[0113] For sub-regions where the ratio of the target prediction value to the corresponding electric energy reserve data is greater than or equal to the second threshold, they can be marked as third sub-regions. Specifically, the third sub-region can be further divided into the third A sub-region and the third B sub-region. The third A sub-region is a sub-region where the ratio of the target prediction value to the corresponding electric energy reserve data is greater than or equal to the second threshold and less than the third threshold. The third A sub-region is a sub-region where the ratio of the target prediction value to the corresponding electric energy reserve data is greater than or equal to the third threshold. The third threshold can be 0.9. In an example, assuming that sub-region j is the third A sub-region, the electricity price for sub-region j can be adjusted according to the following formula: , , is the adjusted electricity price for sub-region j, is the real-time electricity price of sub-region j at the target time; is the target prediction value of sub-region j, is the electric energy reserve data of sub-area j. In another example, assuming that sub-area j is the third sub-area, for sub-area j, the energy storage resources of the first sub-area and / or the second sub-area are called, and the calling amount is ,in is the set of the first sub-region and the second sub-region adjacent to sub-region j, is the idle power data of area k.

[0114] For a detailed example, see Figure 8 , assuming that the area to be predicted includes four sub-areas, the time step is fifteen minutes, the time period to be predicted can be 10:00 to 11:00 am on April 15th, and the time period to be predicted includes four sub-time periods, namely 10:00 to 10:15 am on April 15th, 10:15 to 10:30 am on April 15th, 10:30 to 10:45 am on April 15th, and 10:45 to 11:00 am on April 15th; the first historical time period is 10:00 to 11:00 am on April 14th, and the first historical time period also includes four sub-time periods, namely 10:00 to 10:15 am on April 14th, 10:15 to 10:30 am on April 14th, 10:30 to 10:45 am on April 14th, and 10:45 to 11:00 am on April 14th; the second historical time period is 9:45 to 10:00 am on April 15th. The third historical time period is from 0:00 on April 8 to 23:59 on April 14.

[0115] First, obtain multi-source data (charging timestamp, driving trajectory, charging capacity, actual load, time-of-use electricity price, traffic congestion index, temperature, rainfall, and charging station group usage status) for four sub-time periods in the first historical time period (10:00 am to 11:00 am on April 14). For example, vehicle A moves from area 1 to area 2 at 10:00 am on April 14 to charge, with a charging capacity of 10 kWh; the load in area 1 at 10:00 am on April 14 is 50 kW, and the electricity price is 1.2 yuan / kWh; traffic data shows that the congestion index in area 2 at 10:00 am on April 14 is 6; meteorological data shows that the temperature in area 2 at 10:00 am on April 14 is 25°C, with no rainfall. Aggregate the data of each sub-area every 15 minutes in the first historical time period. For example, the charging capacity of area 2 from 10:00 to 10:15 is 50 kWh, the congestion index is 6, and the temperature is 25°C. Construct a 4*4*9 multidimensional spatiotemporal feature matrix. By inputting the multidimensional spatiotemporal feature matrix into the spatiotemporal coupling prediction model, the initial prediction value of each sub-region can be obtained.

[0116] Then, based on the historical charging transaction data of the past week (from 0:00 on April 8 to 23:59 on April 14), the initial migration probability between sub-areas was calculated. Statistics show that there were 20 vehicles migrating from area 1 to area 2 and 50 vehicles migrating from area 1 to other areas in the past week. The original migration probability is Based on the total number of charging piles in each sub-region, the number of available charging piles, and the difference in electricity prices between sub-regions at the current time (10:00 am on April 15th), the initial migration probability between sub-regions is revised. For example, the number of available charging piles in region 2 is 8 (out of a total of 10), and the electricity price difference between region 2 and region 1 is -0.3 yuan / kWh (region 2 has a lower electricity price). The migration probability is revised. , assuming that after training , ,but , indicating that the probability of users migrating from area 1 to area 2 is adjusted to 0.144 due to the lower electricity price and available charging piles in area 2. According to the revised inter-regional demand migration probability and the initial prediction value of charging demand , generate the revised forecast values ​​for each region in each sub-time period (from 10:00 to 11:00 am on April 15th).

[0117] Then, define the state space ,in is the demand forecast error of the sub-region in the second historical period, is the electricity price fluctuation data, The real-time electricity price, is the average electricity price, is the traffic congestion index, Design reward function for demand overload index; ,in =0.8, =0.3, =0.2 is the weight coefficient. The prediction error of area 2 from 9:45 to 10:00 on April 15th is +5kWh (predicted 60kWh, actual 65kWh). The electricity price fluctuation data at 10:00 on April 15th is +0.1 yuan / kWh. The congestion index is 6, not overloaded, and the status is Calculate Rewards ,The forget gate parameters of the LSTM module are adjusted through the Double DQN algorithm to reduce future prediction errors.

[0118] Finally, based on the results, adjustments were made to the target forecast for Region 2 at 10:00 AM on April 15th, setting it to 62 kWh. A charging demand heat map was generated to visualize the status of each region. Region 2's target forecast was 62 kWh, its energy reserve data was 80 kWh, and its demand accounted for 77.5%. The electricity price remained unchanged. If Region 3's forecast demand was 95 kWh and its capacity was 100 kWh, exceeding 90% of its capacity, the energy storage system would be triggered to discharge, utilizing backup resources from adjacent Region 4 (where the ratio of Region 4's target forecast to its corresponding energy reserve data was less than 0.5).

[0119] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0120] Based on the same inventive concept, the embodiments of the present application also provide a vehicle charging demand migration control device based on multi-source spatiotemporal data fusion for implementing the vehicle charging demand migration control method based on multi-source spatiotemporal data fusion involved above. The implementation solution provided by this device is similar to the implementation solution described in the above method. Therefore, the specific limitations of one or more embodiments of the vehicle charging demand migration control device based on multi-source spatiotemporal data fusion provided below can be referred to the limitations of the vehicle charging demand migration control method based on multi-source spatiotemporal data fusion above, and will not be repeated here.

[0121] In an exemplary embodiment, the present application proposes a vehicle charging demand migration control device based on multi-source spatiotemporal data fusion, the device comprising:

[0122] The first execution module is used to obtain real-time status data of the area to be predicted at a target time, multi-source data in a first historical time period, and demand forecast errors in a second historical time period according to the area to be predicted time period.

[0123] The second execution module is used to perform spatiotemporal alignment processing on the multi-source data according to the preset spatiotemporal units to obtain a multi-dimensional spatiotemporal feature matrix; the multi-dimensional spatiotemporal feature matrix is ​​used to represent the multi-source data of multiple sub-regions of the area to be predicted in multiple sub-historical time periods in the first historical time period.

[0124] The third execution module is used to obtain the target prediction value of each sub-area based on real-time status data, demand prediction error, multi-dimensional spatiotemporal feature matrix and trained spatiotemporal coupling prediction model. The target prediction value is the electric energy demand for vehicle charging in the sub-area during the predicted time period.

[0125] The fourth execution module is used to determine the migration control strategy of the area to be predicted in the time period to be predicted according to each target prediction value.

[0126] Each module in the aforementioned multi-source spatiotemporal data fusion-based vehicle charging demand migration control device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor within a computer device in hardware form, or stored in a computer device memory in software form, allowing the processor to call and execute the corresponding operations of each module.

[0127] In an exemplary embodiment, the present application proposes a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method in any of the above embodiments when executing the computer program.

[0128] The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 9As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless means. The wireless means can be implemented via Wi-Fi, mobile cellular networks, near-field communication (NFC), or other technologies. When executed by the processor, the computer program implements a method for regulating vehicle charging demand migration based on multi-source spatiotemporal data fusion. The display unit of the computer device is used to produce a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0129] Those skilled in the art will understand that Figure 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0130] In an exemplary embodiment, the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method in any of the above embodiments are implemented.

[0131] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0132] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0133] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0134] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A vehicle charging demand migration control method based on multi-source spatiotemporal data fusion, characterized in that: The method comprises: According to the forecast time period of the forecast area, real-time status data of the forecast area at the target time is obtained, as well as multi-source data in the first historical time period and demand forecast errors in the second historical time period; Performing spatiotemporal alignment processing on the multi-source data according to preset spatiotemporal units to obtain a multi-dimensional spatiotemporal feature matrix; the multi-dimensional spatiotemporal feature matrix is ​​used to represent the multi-source data of multiple sub-regions of the area to be predicted in multiple sub-historical time periods in the first historical time period; Obtaining a target prediction value for each of the sub-areas based on the real-time status data, the demand prediction error, the multi-dimensional spatiotemporal feature matrix, and the trained spatiotemporal coupling prediction model, the target prediction value being the electric energy demand for vehicle charging in the sub-area during the time period to be predicted; Determining a migration control strategy for the area to be predicted during the time period to be predicted based on each of the target prediction values; The step of obtaining the target prediction value of each sub-region according to the real-time status data, the demand prediction error, the multi-dimensional spatiotemporal feature matrix, and the trained spatiotemporal coupling prediction model includes: Obtain historical charging transaction data of the area to be predicted within a third historical time period based on the time period to be predicted of the area to be predicted; obtain an initial prediction value for each of the sub-areas based on the multidimensional spatiotemporal feature matrix and the spatiotemporal coupling prediction model; obtain a revised prediction value for each of the sub-areas based on the real-time status data, the historical charging transaction data, and the initial prediction value for each of the sub-areas; and obtain a target prediction value for each of the sub-areas based on each revised prediction value, the real-time status data, and the demand forecast error. The obtaining of a target forecast value for each sub-region according to each of the corrected forecast values, the real-time status data, and the demand forecast error includes: The spatiotemporal coupling prediction model is optimized according to the real-time status data and the demand forecast error; the target prediction value of each sub-area is obtained according to each of the corrected prediction values ​​and the optimized spatiotemporal coupling prediction model; the real-time status data includes the real-time electricity price, demand overload index and traffic congestion index of each sub-area.

2. The method according to claim 1, characterized in that The historical charging transaction data includes the number of charging migrations between multiple sub-areas of the area to be predicted; the historical charging transaction data includes the IDs of all vehicles that charged within the area to be predicted during a third historical time period, the departure area, the charging area, and the charging timestamp of each vehicle's each charging; the number of charging migrations is determined by determining the number of charging migrations between the sub-areas of all vehicles that charged within the area to be predicted during the third historical time period based on the departure area, the charging area, and the charging timestamp of each vehicle's each charging; Migration refers to the act of moving vehicles between regions before charging; The obtaining of a revised predicted value for each sub-area according to the real-time status data, the historical charging transaction data, and the initial predicted value for each sub-area includes: Determine the initial migration probability between each sub-area according to the number of charge migrations between each sub-area; According to the real-time status data, the initial migration probabilities are modified to obtain target migration probabilities; According to the target migration probability and each of the initial prediction values, a revised prediction value of each of the sub-regions is obtained.

3. The method according to claim 2, characterized in that The real-time status data includes the real-time electricity price, the total number of charging piles, and the number of idle charging piles in each sub-area; and the correcting of each initial migration probability based on the real-time status data to obtain a target migration probability includes: Based on the real-time electricity prices of each sub-region, obtain the electricity price difference data between sub-regions; The target migration probability between each sub-area is obtained based on the preset first sensitivity coefficient, the second sensitivity coefficient, the initial migration probabilities, the total number of charging piles and the number of idle charging piles in each sub-area, and the electricity price difference data between each sub-area.

4. The method according to claim 1, wherein The optimizing the spatiotemporal coupling prediction model according to the real-time status data and the demand prediction error includes: Get the average electricity price based on the real-time electricity price of each sub-region; Determining real-time electricity price fluctuation data for each sub-region based on the real-time electricity price of each sub-region and the average electricity price; A reward function for each sub-region is constructed based on a preset first weight coefficient, a second weight coefficient, and a third weight coefficient, as well as the demand forecast error, real-time electricity price fluctuation data, and demand overload index of each sub-region; wherein the first weight coefficient is the weight coefficient of the demand forecast error, the second weight coefficient is the weight coefficient of the demand overload index, and the third weight coefficient is the weight coefficient of the real-time electricity price fluctuation data; The spatiotemporal coupling prediction model is optimized according to the reward function of each sub-area, the demand forecast error, the traffic congestion index, the demand overload index and the real-time electricity price fluctuation data.

5. The method according to claim 1, wherein Determining the migration control strategy of the to-be-predicted area in the to-be-predicted time period according to each of the target prediction values ​​includes: Obtaining the power reserve data of each sub-area in the area to be predicted; When the ratio of the target prediction value of the sub-region to the corresponding electric energy reserve data is less than a first threshold, determining the sub-region as a first sub-region and reducing the real-time electricity price of the first sub-region; When the ratio of the target prediction value of the sub-region to the corresponding electric energy reserve data is greater than or equal to the first threshold and less than the second threshold, the sub-region is determined to be the second sub-region, and the real-time electricity price of the second sub-region is kept unchanged; When the ratio of the target prediction value of the sub-area to the corresponding electric energy reserve data is greater than or equal to a second threshold value, the sub-area is determined to be a third sub-area, and the real-time electricity price of the sub-area is updated according to the target prediction value of the third sub-area and the corresponding electric energy reserve data, and / or energy is dispatched from the first sub-area to the third sub-area according to the target prediction value of the third sub-area and the corresponding electric energy reserve data, as well as the idle electric energy data of the first sub-area; wherein the idle electric energy data is related to the target prediction value of the first sub-area and the corresponding electric energy reserve data.

6. A vehicle charging demand migration control device based on multi-source spatiotemporal data fusion, characterized in that: The device comprises: A first execution module is configured to obtain, based on a to-be-predicted time period of the to-be-predicted area, real-time status data of the to-be-predicted area at a target time, as well as multi-source data within a first historical time period and a demand forecast error within a second historical time period; a second execution module, configured to perform spatiotemporal alignment processing on the multi-source data according to a preset spatiotemporal unit to obtain a multi-dimensional spatiotemporal feature matrix; the multi-dimensional spatiotemporal feature matrix is ​​used to represent the multi-source data of multiple sub-regions of the area to be predicted in multiple sub-historical time periods in the first historical time period; a third execution module, configured to obtain a target prediction value for each of the sub-areas based on the real-time status data, the demand prediction error, the multi-dimensional spatiotemporal feature matrix, and the trained spatiotemporal coupling prediction model, the target prediction value being the electric energy demand for vehicle charging in the sub-area during the time period to be predicted; A fourth execution module is configured to determine a migration control strategy for the area to be predicted in the time period to be predicted according to each of the target prediction values; The third execution module is further configured to obtain historical charging transaction data of the area to be predicted within a third historical time period based on the time period to be predicted of the area to be predicted; obtain an initial prediction value for each of the sub-areas based on the multidimensional spatiotemporal feature matrix and the spatiotemporal coupling prediction model; obtain a revised prediction value for each of the sub-areas based on the real-time status data, the historical charging transaction data, and the initial prediction value for each of the sub-areas; and obtain a target prediction value for each of the sub-areas based on each revised prediction value, the real-time status data, and the demand forecast error. The third execution module is also used to optimize the spatiotemporal coupling prediction model based on the real-time status data and the demand forecast error; obtain the target prediction value of each sub-area based on each of the corrected prediction values ​​and the optimized spatiotemporal coupling prediction model; the real-time status data includes the real-time electricity price, demand overload index and traffic congestion index of each sub-area.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images