Electric vehicle charging resource estimation method, device, equipment and medium

Abstract

The present application relates to the technical field of charge management, and specifically provides a kind of electric vehicle charging resource estimation method, device, equipment and medium, to solve how accurately estimate the charging resource consumption of electric vehicle in different scenarios problem.The method of the present application comprises: obtaining the information related to the charging resource to be estimated, or obtaining the information related to the charging resource to be estimated and the estimated demand, inputting the information related to the charging resource to be estimated or the information related to the charging resource to be estimated and the estimated demand into the trained charging resource estimation model, to obtain the charging resource estimation result, wherein the trained charging resource estimation model comprises a charging resource feature library, a user constant point feature library, a feature extraction module, a semantic information fusion module and a time series feature extraction module.Through the above-mentioned mode, the charging resource consumption in different scenarios can be accurately estimated from time and space based on the trained charging resource estimation model.

Description

Technical Field

[0001] This invention relates to the field of charging management technology, specifically to a method, apparatus, equipment, and medium for estimating electric vehicle charging resources. Background Technology

[0002] With the development of the electric vehicle industry, more and more users are choosing electric vehicles with higher levels of intelligence. The surge in users has also brought about a large demand for charging infrastructure, and the construction of charging resources is in full swing. Against this backdrop, how to build good charging resources and build them in the right way has become an important research topic for new energy vehicle companies and charging infrastructure construction companies.

[0003] Charging resource forecasting is the most crucial element. It can be used to assess the suitability of charging resource locations, aiding in site selection decisions and improving the long-term operational efficiency of charging resource construction companies. Furthermore, it can evaluate the operational status of existing charging resources, helping operators dynamically adjust their marketing strategies. Simultaneously, it can also reflect user demand and the supply-demand relationship of regional charging resources, enhancing the user's charging experience. Therefore, a charging resource forecasting system is vital for construction companies, operators, and users alike.

[0004] Traditional methods for estimating charging resources use expert rules that are variations of economic principles. However, charging resource estimation is related to many factors and the relationships between them are complex. These factors include the competitiveness of specific Points of Interest (POI) locations, the scale of charging stations (number of 7kW and supercharger guns), the competitiveness of the user experience, and the operating status of surrounding charging stations. Expert rules cannot accurately reflect the intrinsic relationships between these factors. At the same time, users' usual locations are scattered across different locations on the map, and the relationship between user demand and charging stations is difficult to quantify using expert rules.

[0005] Accordingly, there is a need in the field for a new method for estimating electric vehicle charging resources to address the above-mentioned problems. Summary of the Invention

[0006] To overcome the above-mentioned shortcomings, this invention is proposed to provide a method, apparatus, device, and medium for predicting electric vehicle charging resources, which solves or at least partially solves the technical problem of how to accurately predict the charging resource usage of electric vehicles in different scenarios.

[0007] In a first aspect, a method for estimating electric vehicle charging resources is provided, the method comprising:

[0008] Obtain information related to the charging resources to be estimated, or obtain information related to the charging resources to be estimated and the estimated demand.

[0009] Input the information related to the charging resources to be estimated or the information related to the charging resources to be estimated and the estimated demand into the trained charging resource estimation model to obtain the charging resource estimation result.

[0010] The trained charging resource prediction model includes a charging resource feature library, a user's permanent location feature library, a feature extraction module, a semantic information fusion module, and a time series feature extraction module.

[0011] In one technical solution of the above-mentioned electric vehicle charging resource estimation method, the information related to the charging resource to be estimated includes at least surrounding charging resource data, user's permanent location data, and the charging resource's own data; the acquisition of the information related to the charging resource to be estimated includes at least:

[0012] The size of the recall circle for the surrounding charging resources is determined based on the surrounding business attributes of the charging resources to be estimated.

[0013] Based on the size of the recall circle, obtain the surrounding charging resource data of the charging resource to be estimated;

[0014] Based on the location of the charging resources to be estimated, obtain the user's permanent location data within a square range of a preset side length.

[0015] In one technical solution of the above-mentioned electric vehicle charging resource estimation method, the estimation scenario includes site selection estimation demand and operation estimation demand. The step of inputting the relevant information of the charging resources to be estimated and the estimation demand into a trained charging resource estimation model to obtain the charging resource estimation result includes:

[0016] When the estimated scenario requirement is a site selection requirement, the site selection requirement and the relevant information of the charging resource to be estimated are input into the trained charging resource estimation model to obtain the site selection estimation result of the charging resource to be estimated.

[0017] When the estimated scenario demand is the estimated business demand, the estimated business demand and the relevant information of the charging resources to be estimated are input into the trained charging resource estimation model to obtain the business estimation result of the charging resources to be estimated.

[0018] In one technical solution of the above-mentioned electric vehicle charging resource estimation method, the method includes training the charging resource estimation model based on at least the following steps:

[0019] Obtain a training sample dataset, wherein the training sample dataset includes a training sample charging resource feature library and a training sample user permanent location feature library.

[0020] The feature extraction module, the semantic information fusion module, and the time series feature extraction module in the charging resource prediction model are trained based on the training sample dataset.

[0021] When the charging resource prediction model converges to a preset error value, the training of the charging resource prediction model is completed.

[0022] In one technical solution of the above-mentioned electric vehicle charging resource estimation method, obtaining the training sample dataset includes:

[0023] The training sample charging resource feature library is constructed based on the profile information of all charging resources in the training samples.

[0024] The feature library of users' permanent locations in the training samples is constructed based on the profile information of all users in the training samples.

[0025] In one technical solution of the above-mentioned electric vehicle charging resource estimation method, the feature extraction module includes a surrounding charging resource feature extraction module, a user demand feature extraction module, and a charging resource-specific feature extraction module. The training of the feature extraction module, the semantic information fusion module, and the time series feature extraction module in the charging resource estimation model based on the training sample dataset includes:

[0026] Based on the surrounding charging resource feature extraction module, the user demand feature extraction module and the charging resource own feature extraction module, feature extraction is performed on the training sample data to obtain the semantic information of surrounding charging resources, the semantic information of user demand and the semantic information of charging resources.

[0027] Based on the semantic information fusion module, the semantic information of the surrounding charging resources, the semantic information of the user demand, and the semantic information of the charging resources themselves are fused to obtain fused semantic information;

[0028] Based on the time series feature extraction module, the fused semantic information is used to extract features to obtain the charging resource prediction result.

[0029] In one technical solution of the above-mentioned electric vehicle charging resource estimation method, the step of extracting features from the training sample data based on the surrounding charging resource feature extraction module, the user demand feature extraction module, and the charging resource own feature extraction module to obtain semantic information of surrounding charging resources, semantic information of user demand, and semantic information of charging resources includes:

[0030] A graph topology is constructed using each of the surrounding charging resources of the charging resource to be evaluated in the training samples as a node; a feature matrix X is constructed based on the features of each surrounding charging resource, and an adjacency matrix A is constructed based on the reciprocal of the distance between the features of each surrounding charging resource; semantic information of the surrounding charging resources is obtained based on the feature matrix X and the adjacency matrix A.

[0031] And / or,

[0032] A demand heat map is constructed based on each of the user's frequently visited locations; features are extracted using convolutional layers, activation is performed using linear rectified layers, and downsampling is performed using pooling layers; semantic information of the user's demand is obtained using fully connected layers.

[0033] And / or,

[0034] Feature encoding is performed on the inherent features of the charging resource to be evaluated in each training sample; the inherent semantic information of the charging resource is obtained based on the fully connected layer.

[0035] In a second aspect, an electric vehicle charging resource estimation device is provided, the device comprising:

[0036] The acquisition module is configured to acquire information related to the charging resources to be estimated, or to acquire the information related to the charging resources to be estimated and the estimated demand.

[0037] The input module is configured to input the information related to the charging resources to be estimated or the information related to the charging resources to be estimated and the estimated demand into a trained charging resource estimation model to obtain the charging resource estimation result.

[0038] In a third aspect, a computer device is provided, the computer device including a processor and a storage device, the storage device being adapted to store a plurality of program codes, the program codes being adapted to be loaded and run by the processor to perform the electric vehicle charging resource estimation method described in any of the above-described technical solutions of the electric vehicle charging resource estimation method.

[0039] In a fourth aspect, a computer-readable storage medium is provided, wherein a plurality of program codes are stored therein, the program codes being adapted to be loaded and run by a processor to perform the electric vehicle charging resource estimation method described in any of the above-described technical solutions.

[0040] The present invention comprises one or more of the following technical solutions:

[0041] Beneficial effects:

[0042] In implementing the technical solution of this invention, firstly, relevant information about the charging resources to be estimated is obtained, or, relevant information about the charging resources to be estimated and the estimated demand are obtained. Then, the relevant information about the charging resources to be estimated or the relevant information about the charging resources to be estimated and the estimated demand are input into a trained charging resource estimation model to obtain the charging resource estimation result. The trained charging resource estimation model includes a charging resource feature library, a user's permanent location feature library, a feature extraction module, a semantic information fusion module, and a time series feature extraction module. Through the above implementation method, based on the trained charging resource estimation model, the charging resource usage in different scenarios can be accurately estimated in both time and space, exhibiting good robustness and estimation accuracy, as well as high engineering application value and generalization ability. Attached Figure Description

[0043] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Furthermore, similar numbers in the drawings are used to denote similar components, wherein:

[0044] Figure 1 This is a schematic flowchart of the main steps of an electric vehicle charging resource estimation method according to an embodiment of the present invention;

[0045] Figure 2 This is a schematic diagram of the main steps for obtaining information related to the charging resources to be estimated according to an embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of the main process of a charging resource estimation method according to an embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the main steps of a charging resource prediction model training method according to an embodiment of the present invention;

[0048] Figure 5 This is a flowchart illustrating the main steps of a method for training a charging resource prediction model, including a feature extraction module, a semantic information fusion module, and a time series feature extraction module, according to an embodiment of the present invention.

[0049] Figure 6 This is a schematic diagram of the surrounding charging resource topology according to an embodiment of the present invention;

[0050] Figure 7 This is a schematic diagram of a user demand thermal pseudo-image according to an embodiment of the present invention;

[0051] Figure 8 This is a schematic diagram of the overall framework of an electric vehicle charging resource estimation method according to an embodiment of the present invention;

[0052] Figure 9 This is a main structural block diagram of an electric vehicle charging resource estimation device according to an embodiment of the present invention;

[0053] Figure 10 This is a schematic diagram of the main structure of an electronic device according to an embodiment of the present invention.

[0054] List of reference numerals in the attached diagram:

[0055] 901: Acquisition module; 902: Input module; 1001: Processor; 1002: Storage device. Detailed Implementation

[0056] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0057] In the description of this invention, "module" and "processor" can include hardware, software, or a combination of both. A module can include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. A processor can be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor can be implemented in software, in hardware, or a combination of both. Non-transitory computer-readable storage media includes any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B. The terms "at least one A or B" or "at least one of A and B" have a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "a" or "this" can also include plural forms.

[0058] Traditional methods for estimating charging resources rely on expert rules derived from economic principles. However, charging resource estimation is influenced by numerous and complex factors, including the competitiveness of charging resource POI locations, station size (number of 7kW and supercharger guns), user experience competitiveness, and the operational status of surrounding charging stations. Expert rules cannot accurately reflect the intrinsic relationships between these factors. Furthermore, users' usual locations are scattered across different areas on a map, making it difficult to quantify the relationship between user demand and charging stations using expert rules. To address these issues, this invention provides a method, apparatus, device, and medium for estimating electric vehicle charging resources.

[0059] See appendix Figure 1 , Figure 1 This is a schematic flowchart illustrating the main steps of an electric vehicle charging resource estimation method according to an embodiment of the present invention. Figure 1 As shown, the electric vehicle charging resource estimation method in this embodiment of the invention mainly includes the following steps S101 to S102.

[0060] Step S101: Obtain information related to the charging resources to be estimated, or obtain the information related to the charging resources to be estimated and the estimated demand.

[0061] Step S102: Input the relevant information of the charging resources to be estimated or the relevant information of the charging resources to be estimated and the estimated demand into the trained charging resource estimation model to obtain the charging resource estimation result.

[0062] In some implementations, the trained charging resource prediction model includes a charging resource feature library, a user's frequently visited location feature library, a feature extraction module, a semantic information fusion module, and a time series feature extraction module.

[0063] The method described in steps S101 to S102 above can accurately predict the charging resource usage in different scenarios in terms of time and space based on the trained charging resource prediction model. It has good robustness and prediction accuracy, and also has high engineering application value and generalization.

[0064] The following provides a further explanation of steps S101 to S102.

[0065] In some embodiments of step S101 above, the information related to the charging resources to be estimated includes at least surrounding charging resource data, user's permanent location data, and the charging resources' own data to be estimated.

[0066] Further, please refer to the appendix. Figure 2 , Figure 2 This is a schematic flowchart illustrating the main steps of obtaining information related to the charging resources to be estimated according to an embodiment of the present invention. Figure 2 As shown, in step S101, obtaining information related to the charging resources to be estimated includes at least the following steps S1011 to S1013.

[0067] Step S1011: Determine the size of the recall circle for the surrounding charging resources based on the surrounding business attributes of the charging resources to be estimated.

[0068] In some implementations, the surrounding business attributes of the charging resources to be estimated include POI business attributes, with each POI business attribute corresponding to a different recall circle size.

[0069] Among them, POI stands for Point of Interest, which refers to the specific information of each location point in the map data, including name, category, longitude and latitude, and information about nearby shops.

[0070] Step S1012: Obtain surrounding charging resource data of the charging resource to be estimated based on the size of the recall circle.

[0071] In some implementations, the surrounding charging resource data of the charging resource to be estimated includes the name, category, longitude and latitude of the surrounding charging resource.

[0072] Step S1013: Obtain user location data within a square area with a preset side length based on the location of the charging resources to be estimated.

[0073] It should be noted that the side length of the square is set by those skilled in the art according to actual needs in practical applications, and is not limited here.

[0074] In some implementations, after acquiring surrounding charging resource data, user's resident location data, and the data of the charging resources to be estimated, these data can be combined to form three sequences. When estimating charging resources, the estimation result is obtained by dynamically sliding a time window through the historical sequence information of surrounding charging resources, the historical sequence information of user's resident locations, and the historical sequence information of the charging resources themselves. For example, by dynamically sliding the three historical sequence information from the previous week, the estimation result of charging resources for the next two days can be obtained.

[0075] The above is a further explanation of step S101. The following is a further explanation of step S102.

[0076] In some implementations of step S102 above, the estimated scenarios include site selection estimated demand or operation estimated demand. Further, the relevant information of the charging resources to be estimated and the estimated demand are input into the trained charging resource estimation model to obtain the charging resource estimation result.

[0077] In some implementations, when the estimated scenario demand is a site selection demand, the site selection demand and the relevant information of the charging resources to be estimated are input into the trained charging resource estimation model to obtain the site selection estimation result of the charging resources to be estimated.

[0078] Specifically, when the user inputs a site selection prediction scenario, the system will display the required input of different power gun numbers, latitude and longitude locations, and POI business type attributes of potential charging resources. The charging resource prediction model will then use the worst-performing combination of operating characteristics (such as the worst charging success rate) and the best-performing combination of operating characteristics to infer and calculate the best and worst prediction results for the site to be predicted.

[0079] When the estimated scenario demand is the same as the estimated business demand, the relevant information of the charging resources to be estimated and the experience-based estimated demand are input into the trained charging resource estimation model to obtain the business estimation result.

[0080] Specifically, when the user inputs an estimated scenario requirement of business estimation, the system displays the name of the charging resource and the potential price range of the charging resource. The business features of the charging resource itself, obtained by the charging resource estimation model, are derived from the business features of the charging resource in the charging resource feature library, and the system displays the estimated results of different prices within the price range to the user.

[0081] In some implementations, see Appendix Figure 3 , Figure 3 This is a schematic diagram of the main process of a charging resource estimation method according to an embodiment of the present invention. Figure 3 As shown, when estimating charging resources, the user first inputs the estimated scenario requirements. Then, based on the estimated scenario requirements, the user inputs relevant information about the charging resources to be estimated. After feature extraction based on the charging resource feature library and the user's permanent location feature library, the user performs relevant inference calculations through the charging resource estimation model to obtain the corresponding charging resource estimation results.

[0082] Further, please refer to the appendix. Figure 4 , Figure 4 This is a schematic flowchart illustrating the main steps of a charging resource prediction model training method according to an embodiment of the present invention. Figure 4 As shown, the method for training the charging resource estimation model in this embodiment of the invention mainly includes the following steps S401 to S403.

[0083] Step S401: Obtain the training sample dataset.

[0084] The training sample dataset includes a training sample charging resource feature library and a training sample user's permanent location feature library.

[0085] In some implementations, a training sample charging resource feature library can be constructed based on the profile information of all charging resources in the training sample; and a training sample user resident location feature library can be constructed based on the profile information of all users in the training sample.

[0086] Furthermore, the profile information for all charging resources includes the latitude and longitude location of all charging resources, the number of charging guns of different power types, the daily number of charging orders for different power types of charging guns, parking space type, peak-valley-average price, charging success rate, charging station occupancy rate, actual charging speed of different power types of charging guns, occupancy rate of different power types of charging guns, and POI business type, among other characteristics. Among them, the power type is divided into three types: slow, medium, and fast, based on the rated charging power of the charging gun.

[0087] All user profile information includes the latitude and longitude of the user's usual location, the Cartesian coordinates of the user's usual location, and the number of charging orders the user has placed in the past 180 days.

[0088] Step S402: Train the feature extraction module, semantic information fusion module, and time series feature extraction module in the charging resource prediction model based on the training sample dataset.

[0089] In some implementations, the feature extraction module includes a surrounding charging resource feature extraction module, a user demand feature extraction module, and a charging resource own feature extraction module.

[0090] Further, please refer to the appendix. Figure 5 , Figure 5 This is a schematic flowchart illustrating the main steps of a method for training a charging resource prediction model, comprising a feature extraction module, a semantic information fusion module, and a time series feature extraction module, according to an embodiment of the present invention. Figure 5 As shown, step S402 includes at least the following steps S4021 to S4023.

[0091] Step S4021: Based on the surrounding charging resource feature extraction module, user demand feature extraction module and charging resource own feature extraction module, feature extraction is performed on the training sample data to obtain the semantic information of surrounding charging resources, semantic information of user demand and semantic information of charging resources.

[0092] Furthermore, in some implementations, obtaining semantic information about surrounding charging resources based on the surrounding charging resource feature extraction module includes:

[0093] (1) Construct a graph topology with each surrounding charging resource of the charging resource to be evaluated in the training sample as a node;

[0094] (2) Construct a feature matrix X based on the characteristics of each surrounding charging resource, and construct an adjacency matrix A based on the reciprocal of the distance between each surrounding charging resource feature;

[0095] (3) Obtain semantic information of surrounding charging resources based on feature matrix X and adjacency matrix A.

[0096] See appendix Figure 6 , Figure 6 This is a schematic diagram of the surrounding charging resource topology according to an embodiment of the present invention. Figure 6 As shown, a non-Euclidean graph topology can be constructed using each surrounding charging resource as a node.

[0097] The feature matrix X is composed of the characteristics of each surrounding charging resource. The elements in the feature matrix X include the daily number of charging orders per gun for different power types of surrounding charging resources, the normalized distance to the charging resource to be estimated, the number of guns of different power types, parking space type, peak-valley-flat price, charging success rate, charging station occupancy rate, actual charging speed of guns of different power types, occupancy rate of guns of different power types, and POI business type, etc.

[0098] The normalized distance to the charging resource to be estimated is obtained by calculating the distance between each charging resource to be estimated and the surrounding charging resources and dividing it by the recall radius.

[0099] Furthermore, the adjacency matrix A is formed by representing the relationship between each surrounding charging resource using the reciprocal of the distance between them. It should be noted that when two charging resources exceed a preset distance, the elements of adjacency matrix A become 0. Those skilled in the art can set the preset distance according to actual needs in practical applications; no limitation is made here.

[0100] The feature extraction model for surrounding charging resources consists of a two-layer graph convolutional neural network (GCN) and a fully connected network. The feature matrix X and adjacency matrix A of the surrounding charging resources are calculated by the feature extraction model to obtain the topological semantic information α of the surrounding charging resources.

[0101] It should be noted that the peripheral charging resource feature extraction module in the above method, which consists of a two-layer graph convolutional neural network (GCN) and a fully connected network connected in series, is only an illustrative example. The graph convolutional layer can be replaced with other graph convolutional layers of various forms, including but not limited to graph attention networks, inductive learning (GraphSAGE), etc. As long as it does not violate the technical concept of the present invention, those skilled in the art can choose according to actual needs in practical applications, and no limitation is made here.

[0102] The above is the relevant content regarding the semantic information of surrounding charging resources obtained based on the surrounding charging resource feature extraction module.

[0103] In some implementations, the semantic information of user needs obtained based on the user need feature extraction module includes:

[0104] (1) Construct a demand heat map based on each user's permanent location;

[0105] (2) Extract features based on convolutional layers, perform activation based on linear rectified layers, and perform downsampling based on pooling layers;

[0106] (3) Obtain semantic information of user needs based on the fully connected layer.

[0107] For details, please refer to the appendix. Figure 7 , Figure 7This is a schematic diagram of a user demand thermal pseudo-image according to an embodiment of the present invention. Figure 7 As shown, the size of each pixel in the required thermal pseudo-image constructed based on each user's permanent location is ε, centered on the pixel position of each user's permanent location. The thermal value generated by each user's permanent location on the thermal pseudo-image is:

[0108]

[0109] Where x and y are the Cartesian coordinates of the user's permanent location, x0 and y0 are the Cartesian coordinates of the user's permanent location relative to the charging resources to be estimated, and σ p ζ represents the number of charging orders placed by users at resident locations over the past 180 days.

[0110] Furthermore, the heat values ​​generated by each user's permanent location are superimposed to form a pseudo-thermal image.

[0111] The user demand feature extraction model consists of a fully connected layer with one convolutional layer, two linear rectified ReLU layers, one pooling layer, and a thermal pseudo-image. Features are extracted from the thermal pseudo-image through the convolutional layer, activated by the ReLU layer, and then input to the pooling layer for downsampling. Finally, the topological semantic information β of the user demand is obtained through the fully connected layer.

[0112] It should be noted that in the above method, the user demand feature extraction module consists of a fully connected layer with one convolutional layer, two linear rectified ReLU layers, and one pooling layer, which is only for illustrative purposes. The convolutional layer can be replaced with various forms of convolutional layers, including but not limited to grouped convolution, separable convolution, dilated convolution, depthwise convolution, etc. As long as it does not violate the technical concept of the present invention, those skilled in the art can choose according to actual needs in practical applications, and no limitation is made here.

[0113] The above is the relevant content regarding obtaining semantic information about user needs based on the user needs feature extraction module.

[0114] In some implementations, obtaining the semantic information of charging resources based on the charging resource-specific feature extraction module includes:

[0115] (1) Encode the features of the charging resources to be evaluated in each training sample;

[0116] (2) Obtain the inherent semantic information of charging resources based on the fully connected layer.

[0117] Specifically, when encoding the inherent characteristics of charging resources, these characteristics include the number of charging guns of different power types, parking space type, peak-valley-flat price, charging success rate, charging station occupancy rate, actual charging speed of charging guns of different power types, occupancy rate of charging guns of different power types, and POI business type.

[0118] The model within the charging resource feature extraction module consists of a fully connected layer, after which the semantic information γ of the charging resource can be obtained.

[0119] The above is a further explanation of step S4021. The following is a further explanation of step S4022.

[0120] Step S4022: Based on the semantic information fusion module, perform semantic information fusion on the semantic information of surrounding charging resources, user demand semantic information and the semantic information of the charging resources themselves to obtain fused semantic information.

[0121] In some implementations, the semantic information fusion module concatenates the semantic information α of surrounding charging resources, the semantic information β of user demand, and the semantic information γ of the charging resources themselves to obtain fused semantic information λ = [α, β, γ].

[0122] Step S4023: Based on the time series feature extraction module, perform feature extraction on the fused semantic information λ to obtain the charging resource prediction result.

[0123] In some implementations, the time-series feature extraction module extracts features from the sequence points of the semantic information α of surrounding charging resources, the semantic information β of user demand, the semantic information γ of charging resources themselves, and the fused semantic information λ, to obtain the fused semantic information sequence Ψ = [λ1,…,λ]. n-1 ,λ n ], where n is the length of the sequence.

[0124] Furthermore, based on a two-layer recurrent neural network (RNN), features are extracted from the fused semantic information sequence. Finally, the usage of charging resources under three power types (slow, medium, and fast) is estimated through a fully connected layer with three output nodes.

[0125] It should be noted that the use of RNN network layers in the time series feature extraction module in the above method is only illustrative. The RNN network layer can be replaced by network units such as Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Bi-LSTM, and Bi-RNN. Layers to avoid overfitting can be added to the network architecture, including but not limited to regularization and dropout (dropout refers to temporarily discarding a portion of neural network units from the network with a certain probability during the training process of a deep learning network, which is equivalent to finding a leaner network from the original network). As long as it does not violate the technical concept of the present invention, those skilled in the art can choose according to actual needs in practical applications, and no limitation is made here.

[0126] After completing step S4023, step S402 is completed. The following is a further explanation of step S403.

[0127] Step S403: When the charging resource prediction model converges to the preset error value, the training of the charging resource prediction model is completed.

[0128] In some implementations, the feature extraction module, semantic information fusion module, and time series feature extraction module are trained end-to-end based on the training sample dataset until the model calculation converges to the minimum error value. Finally, the model with the optimal weight parameters can be saved to a specified file so that when performing charging resource estimation, the model file with the optimal weight parameters can be loaded for charging resource estimation based on the user's input of relevant information about the charging resources to be estimated and / or the estimation requirements.

[0129] The above-described method for training the charging resource prediction model employs the error backpropagation gradient descent method, where the loss function is the root mean square error function. It should be noted that, as long as it does not deviate from the technical concept of this invention, those skilled in the art can choose the training method according to actual needs in practical applications; no limitations are imposed here.

[0130] After training the model, charging resources can be estimated based on the model and user needs. (See appendix.) Figure 8 , Figure 8 This is a schematic diagram of the overall framework of an electric vehicle charging resource estimation method according to an embodiment of the present invention. Figure 8As shown, based on the surrounding charging resource feature extraction module, user demand feature extraction module, and charging resource own feature extraction module, features are extracted from the charging resource feature library and the user's permanent location feature library. Then, semantic fusion and feature extraction are performed based on the semantic information fusion module and the time series feature extraction module to complete the training of the charging resource prediction model. Furthermore, when a user has a prediction need, the user inputs the information of the charging resource to be predicted and / or the prediction need, and the charging resource prediction model obtains the charging resource prediction result based on the input information.

[0131] It should be noted that although the steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effects of the present invention, different steps do not necessarily have to be executed in such an order. They can be executed simultaneously (in parallel) or in other orders, and these variations are all within the scope of protection of the present invention.

[0132] Furthermore, the present invention also provides an electric vehicle charging resource estimation device. (See appendix) Figure 9 , Figure 9 This is a main structural block diagram of an electric vehicle charging resource estimation device according to an embodiment of the present invention. Figure 9 As shown, the electric vehicle charging resource estimation device in this embodiment of the invention mainly includes an acquisition module 901 and an input module 902. In some embodiments, the acquisition module 901 can be configured to acquire information related to the charging resource to be estimated, or to acquire the information related to the charging resource to be estimated and the estimated demand. The input module 902 can be configured to input the information related to the charging resource to be estimated or the information related to the charging resource to be estimated and the estimated demand into a trained charging resource estimation model to obtain the charging resource estimation result. In one embodiment, a description of the specific functions can be found in steps S101 to S102.

[0133] Furthermore, it should be understood that since the various modules are only provided to illustrate the functional units of the device of the present invention, the physical devices corresponding to these modules may be the processor itself, or a part of the processor's software, hardware, or a combination of software and hardware. Therefore, the number of modules shown in the figures is merely illustrative.

[0134] Those skilled in the art will understand that the various modules in the device can be adaptively split or combined. Such splitting or combining of specific modules will not cause the technical solution to deviate from the principles of the present invention; therefore, the technical solutions after splitting or combining will fall within the protection scope of the present invention.

[0135] The aforementioned electric vehicle charging resource estimation device is used for execution Figure 1The electric vehicle charging resource estimation method embodiments shown are similar in technical principle, the technical problems solved and the technical effects produced. Those skilled in the art can clearly understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the electric vehicle charging resource estimation device can be referred to the content described in the embodiments of the electric vehicle charging resource estimation method, and will not be repeated here.

[0136] Those skilled in the art will understand that all or part of the processes in the method of the above embodiment of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0137] Furthermore, the present invention also provides an electronic device. (See appendix.) Figure 10 , Figure 10 This is a schematic diagram of the main structure of an electronic device according to an embodiment of the present invention. Figure 10 As shown, the electronic device in this embodiment of the invention mainly includes a processor 1001 and a storage device 1002. The storage device 1002 can be configured to store a program for executing the electric vehicle charging resource estimation method of the above-described method embodiment. The processor 1001 can be configured to execute the program in the storage device 1002, which includes, but is not limited to, the program for executing the electric vehicle charging resource estimation method of the above-described method embodiment. For ease of explanation, only the parts related to the embodiments of the present invention are shown. For specific technical details not disclosed, please refer to the method section of the embodiments of the present invention.

[0138] In some possible embodiments of the present invention, the electronic device may include multiple processors 1001 and multiple storage devices 1002. The program executing the electric vehicle charging resource estimation method of the above-described method embodiments can be divided into multiple subroutines. Each subroutine can be loaded and run by a processor 1001 to execute different steps of the electric vehicle charging resource estimation method of the above-described method embodiments. Specifically, each subroutine can be stored in different storage devices 1002, and each processor 1001 can be configured to execute programs in one or more storage devices 1002 to jointly implement the electric vehicle charging resource estimation method of the above-described method embodiments. That is, each processor 1001 executes different steps of the electric vehicle charging resource estimation method of the above-described method embodiments to jointly implement the electric vehicle charging resource estimation method of the above-described method embodiments.

[0139] The aforementioned multiple processors 1001 can be processors deployed on the same device. For example, the aforementioned electronic device can be a high-performance device composed of multiple processors, and the aforementioned multiple processors 1001 can be processors configured on that high-performance device. Alternatively, the aforementioned multiple processors 1001 can also be processors deployed on different devices. For example, the aforementioned electronic device can be a server cluster, and the aforementioned multiple processors 1001 can be processors on different servers within the server cluster; the aforementioned electronic device can be a driving equipment cluster, and the aforementioned multiple processors 1001 can be processors on different driving devices within the driving equipment cluster.

[0140] Furthermore, the present invention also provides a computer-readable storage medium. In one embodiment of the computer-readable storage medium according to the present invention, the computer-readable storage medium can be configured to store a program for executing the electric vehicle charging resource estimation method of the above-described method embodiments. This program can be loaded and run by a processor to implement the above-described electric vehicle charging resource estimation method. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The computer-readable storage medium can be a storage device comprising various electronic devices. Optionally, in the embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0141] The technical solution of the present invention has been described above with reference to one embodiment shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions resulting from such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for estimating charging resources for electric vehicles, characterized in that, The method includes: Obtain information related to the charging resources to be estimated, or obtain information related to the charging resources to be estimated and the estimated demand, wherein the information related to the charging resources to be estimated includes surrounding charging resource data, user's permanent location data and the charging resources' own data. Input the information related to the charging resources to be estimated or the information related to the charging resources to be estimated and the estimated demand into the trained charging resource estimation model to obtain the charging resource estimation result. The charging resource prediction model includes a feature extraction module, a semantic information fusion module, and a time series feature extraction module. The feature extraction module includes a surrounding charging resource feature extraction module, a user demand feature extraction module, and a charging resource own feature extraction module. The step of inputting the information related to the charging resources to be estimated, or the information related to the charging resources to be estimated and the estimated demand, into a trained charging resource estimation model to obtain the charging resource estimation result includes: The surrounding charging resource feature extraction module extracts features from the information related to the charging resources to be estimated to obtain semantic information of the surrounding charging resources. This includes: constructing a graph topology with each surrounding charging resource in the surrounding charging resource data as a node; constructing a feature matrix X based on the features of each surrounding charging resource; constructing an adjacency matrix A based on the reciprocal of the distance between the features of each surrounding charging resource; and obtaining the semantic information of the surrounding charging resources based on the feature matrix X and the adjacency matrix A. Based on the user demand feature extraction module, feature extraction is performed on the charging resource information to be estimated to obtain user demand semantic information, including: constructing a demand heat map pseudo image based on each user's permanent location, extracting features based on convolutional layers, performing activation based on linear rectifier layers, performing downsampling based on pooling layers, and obtaining the user demand semantic information based on fully connected layers. Based on the charging resource-specific feature extraction module, feature extraction is performed on the relevant information of the charging resource to be estimated to obtain the charging resource-specific semantic information, including: performing feature encoding on the charging resource-specific data to be estimated, and obtaining the charging resource-specific semantic information based on the fully connected layer; Based on the semantic information fusion module, the semantic information of the surrounding charging resources, the semantic information of the user demand, and the semantic information of the charging resources themselves are fused to obtain fused semantic information; Based on the time series feature extraction module, the fused semantic information is used to extract features to obtain the charging resource prediction result.

2. The method for estimating electric vehicle charging resources according to claim 1, characterized in that, The acquisition of information related to the charging resources to be estimated includes at least: The size of the recall circle for the surrounding charging resources is determined based on the surrounding business attributes of the charging resources to be estimated. Based on the size of the recall circle, obtain the surrounding charging resource data of the charging resource to be estimated; Based on the location of the charging resources to be estimated, obtain the user's permanent location data within a square range of a preset side length.

3. The method for estimating electric vehicle charging resources according to claim 1, characterized in that, The estimated scenarios include site selection estimation needs and operational estimation needs. The step of inputting the relevant information of the charging resources to be estimated and the estimated needs into a trained charging resource estimation model to obtain the charging resource estimation results includes: When the estimated scenario requirement is a site selection requirement, the site selection requirement and the relevant information of the charging resource to be estimated are input into the trained charging resource estimation model to obtain the site selection estimation result of the charging resource to be estimated. When the estimated scenario demand is the estimated business demand, the estimated business demand and the relevant information of the charging resources to be estimated are input into the trained charging resource estimation model to obtain the business estimation result of the charging resources to be estimated.

4. The method for estimating electric vehicle charging resources according to claim 1, characterized in that, The method includes training the charging resource estimation model based on at least the following steps: Obtain a training sample dataset, wherein the training sample dataset includes a training sample charging resource feature library and a training sample user permanent location feature library. The feature extraction module, the semantic information fusion module, and the time series feature extraction module in the charging resource prediction model are trained based on the training sample dataset. When the charging resource prediction model converges to a preset error value, the training of the charging resource prediction model is completed.

5. The method for estimating electric vehicle charging resources according to claim 4, characterized in that, The training sample dataset obtained includes: The training sample charging resource feature library is constructed based on the profile information of all charging resources in the training samples. The feature library of users' permanent locations in the training samples is constructed based on the profile information of all users in the training samples.

6. An electric vehicle charging resource estimation device, characterized in that, The apparatus is used to perform the electric vehicle charging resource estimation method according to any one of claims 1 to 5, the apparatus comprising: The acquisition module is configured to acquire information related to the charging resources to be estimated, or to acquire the information related to the charging resources to be estimated and the estimated demand. The input module is configured to input the information related to the charging resources to be estimated or the information related to the charging resources to be estimated and the estimated demand into a trained charging resource estimation model to obtain the charging resource estimation result.

7. An electronic device comprising a processor and a storage device, said storage device being adapted to store a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by the processor to perform the electric vehicle charging resource estimation method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to perform the electric vehicle charging resource estimation method according to any one of claims 1 to 5.

Images