A method, system, and medium for evaluating the response of charging stations based on spatiotemporal traffic correlation.

Abstract

This invention relates to the field of power system data processing technology, and particularly to a method, system, and medium for evaluating the response of charging stations based on spatiotemporal traffic flow correlation. The method includes: acquiring road traffic monitoring data, charging station operation data, and station resource status data, and constructing a spatiotemporal graph dataset; inputting the spatiotemporal graph dataset into a graph convolutional network for spatial feature learning to obtain a spatial feature sequence; inputting the spatial feature sequence into a long short-term memory network for time-dependency modeling to obtain a traffic flow prediction result for the target time period; performing joint distribution fitting between the traffic flow prediction result and historical charging demand in the charging station operation data to obtain a conditional charging demand distribution for the target time period; and calculating the response potential index of each charging station for the target time period based on the conditional charging demand distribution and station resource status data. This invention effectively solves the problem that existing technologies struggle to accurately assess the response potential of electric vehicle charging stations by incorporating traffic flow changes.

Description

Technical Field

[0001] This invention relates to the field of power system data processing technology, and in particular to a method, system and medium for evaluating the response of charging stations based on spatiotemporal flow correlation. Background Technology

[0002] Electric vehicle charging stations are crucial infrastructure for providing power to electric vehicles. Their operational status directly impacts vehicle charging efficiency, user service experience, and the utilization rate of power distribution resources. The station's response potential typically refers to its ability to handle, regulate, and serve charging demand under specific traffic and load conditions. With the continuous growth of electric vehicle ownership and the ongoing improvement of charging infrastructure, charging demand is exhibiting greater dynamism, regionality, and volatility, and is particularly susceptible to changes in road traffic conditions.

[0003] In existing technologies, solutions for charging station layout, load analysis, and resource scheduling are mostly based on fixed geographical locations, historical operating data, or relatively static characteristic information for evaluation and decision-making. Some methods have begun to incorporate factors such as traffic flow and load curves to assist in the analysis, but overall, they still rely mainly on static analysis or single-feature analysis. This makes it difficult to comprehensively reflect the coupled characteristics of traffic flow in spatial distribution and temporal changes, and also makes it difficult to accurately depict the complex relationship between changes in traffic flow and changes in charging demand. As a result, existing technologies often struggle to accurately determine the load capacity, response speed, and demand fulfillment level of different charging stations when facing scenarios with significant traffic fluctuations and frequent demand changes. This leads to problems such as inadequate allocation of charging resources, insufficient scheduling flexibility, and reduced service capacity during peak hours. Therefore, how to accurately assess the response potential of electric vehicle charging stations in conjunction with traffic flow changes has become a pressing technical problem that needs to be solved in this field.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the general background of this disclosure and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a method, system, and medium for evaluating the response of charging stations based on spatiotemporal traffic correlation, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A charging station response evaluation method based on spatiotemporal traffic correlation, the method comprising: Acquire road traffic monitoring data, charging station operation data, and station resource status data, and construct a spatiotemporal map dataset representing the correlation between road areas based on the road traffic monitoring data; The spatiotemporal graph dataset is input into a graph convolutional network for spatial feature learning to obtain a spatial feature sequence representing the correlation of traffic flow in different road areas; The spatial feature sequence is input into a long short-term memory network for time-dependent modeling to obtain the traffic flow prediction result for the target time period; Based on the Copula dependency model, the traffic flow prediction results are jointly fitted with the historical charging demand in the charging station operation data to obtain the conditional charging demand distribution for the target time period. Based on the conditional charging demand distribution and the site resource status data, the response potential index of each charging site during the target time period is calculated, and the corresponding site response potential assessment result is output.

[0007] Furthermore, a spatiotemporal graph dataset representing the relationships between road regions is constructed, including: The missing values ​​in the road traffic monitoring data are processed and replaced with 0. The road traffic monitoring data after missing value processing is used to generate sequence samples, and the data of the next moment is predicted using the first 50 time series data. The road traffic monitoring data generated from the sequence samples is normalized to obtain the input data used to construct the spatiotemporal graph dataset.

[0008] Furthermore, constructing a spatiotemporal graph dataset representing the relationships between road regions also includes: The road network is abstracted as a graph structure, with road segments as nodes and the connection relationships between road segments as edges; Construct the adjacency matrix corresponding to the graph structure based on geographical location relationships and road connectivity relationships; The nodes are assigned current traffic flow, average vehicle speed, and occupancy characteristics to form the spatiotemporal graph dataset.

[0009] Furthermore, spatial feature sequences are learned, including: The spatiotemporal graph dataset is input into the graph convolutional network; Convolutional aggregation is performed on the features of the neighboring nodes of each node based on the adjacency matrix; The spatial feature sequence is obtained by generating node spatial features that incorporate neighborhood information through a learnable weight matrix and a nonlinear activation function.

[0010] Furthermore, the traffic flow prediction results for the target time period are obtained, including: The spatial feature sequence is input into the first layer of the long short-term memory network, and the intermediate states at each time step are retained; The output of the first layer of the Long Short-Term Memory network is input into the second layer of the Long Short-Term Memory network, and the hidden state of the last time step is output. The hidden layer data output by the Long Short-Term Memory network is input into the fully connected layer; The traffic flow prediction result for the target time period is obtained based on the fully connected layer.

[0011] Furthermore, a joint distribution fitting is performed on historical charging demand, including: A correlation dataset is constructed based on the traffic flow prediction results and historical data of charging stations, wherein the correlation dataset includes traffic flow and corresponding charging demand at each time point; The maximum likelihood estimation method is used to fit the marginal distribution of the traffic flow and the charging demand, respectively. The joint probability distribution of the traffic flow and the charging demand is constructed based on the Gaussian Copula function, and the charging demand is predicted given the traffic flow to obtain the conditional charging demand distribution.

[0012] Furthermore, the response potential indicators of each charging station during the target time period are obtained, including: The charging demand of the charging station is obtained based on the charging demand distribution under the aforementioned conditions. Predict the demand response volume of the charging station based on the charging demand and the station resource status data; The load capacity and response speed of each charging station under different traffic flow conditions are evaluated based on the demand response volume to determine the degree to which charging demand is met and to form the evaluation result of the station response potential.

[0013] Furthermore, the corresponding site response potential assessment results are output, including: Based on the assessment results of the station's response potential, identify the peak traffic periods corresponding to the target time period; According to the aforementioned peak traffic periods, charging stations with loads below average charging conditions will be prioritized for scheduling. By dynamically adjusting the working status of charging piles, overload can be avoided, thereby optimizing the allocation of charging resources.

[0014] A charging station response evaluation system based on spatiotemporal traffic correlation, the system comprising: The data acquisition module acquires road traffic monitoring data, charging station operation data, and station resource status data, and constructs a spatiotemporal map dataset representing the correlation between road areas based on the road traffic monitoring data; The feature learning module inputs the spatiotemporal graph dataset into the graph convolutional network to learn spatial features and obtain spatial feature sequences that characterize the correlation of traffic flow in different road areas. The traffic flow prediction module inputs the spatial feature sequence into a long short-term memory network for time-dependent modeling to obtain the traffic flow prediction results for the target time period. The distribution fitting module uses the Copula dependency model to perform joint distribution fitting between the traffic flow prediction results and the historical charging demand in the charging station operation data to obtain the conditional charging demand distribution for the target time period. The potential assessment module calculates the response potential index of each charging station during the target time period based on the distribution of conditional charging demand and the status data of station resources, and outputs the corresponding station response potential assessment results.

[0015] A computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, can implement the aforementioned charging station response evaluation method based on spatiotemporal traffic correlation.

[0016] The technical solution of this invention can achieve the following technical effects: By constructing a road traffic map and using graph convolutional networks and long short-term memory networks to predict traffic flow in the time and space, and then using the Copula model to establish the correlation between traffic flow and charging demand, the response potential assessment and resource scheduling optimization of electric vehicle charging stations can be realized. This effectively solves the problem that existing technologies are unable to accurately assess the response potential of electric vehicle charging stations and achieve efficient resource scheduling under the conditions of spatiotemporal fluctuations in traffic flow and nonlinear correlation of charging demand.

[0017] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the charging station response evaluation method based on spatiotemporal traffic correlation. Figure 2 A flowchart illustrating the modeling process of the method established in this scheme; Figure 3 The results are for predicting traffic flow; Figure 4 The graph illustrates the correlation between traffic flow and charging demand using a Gaussian Copula function. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] Example 1; like Figure 1 As shown, this application provides a charging station response evaluation method based on spatiotemporal traffic correlation, the method including: S10: Acquire road traffic monitoring data, charging station operation data, and station resource status data, and construct a spatiotemporal map dataset representing the correlation between road areas based on the road traffic monitoring data; S20: Input the spatiotemporal graph dataset into the graph convolutional network to learn spatial features and obtain spatial feature sequences that characterize the correlation of traffic flow in different road areas; S30: Input the spatial feature sequence into the long short-term memory network for time-dependent modeling to obtain the traffic flow prediction results for the target time period; S40: Based on the Copula dependency model, the distribution of the predicted traffic flow and the historical charging demand in the charging station operation data are jointly fitted to obtain the conditional charging demand distribution for the target period. S50: Calculates the response potential index of each charging station during the target time period based on the distribution of conditional charging demand and the status data of station resources, and outputs the corresponding station response potential assessment results.

[0023] Specifically, this embodiment provides a method for evaluating the response potential of electric vehicle charging stations based on spatiotemporal correlation learning. First, road traffic monitoring data, charging station operation data, and station resource status data are acquired. Road traffic monitoring data characterizes the traffic operation status of the road area, charging station operation data characterizes historical charging demand or load changes, and station resource status data characterizes the schedulable resource status of each charging station. Based on the road traffic monitoring data, a spatiotemporal graph dataset representing the correlation relationships of the road area is constructed. Specifically, the road network is abstracted into a graph structure, road segments are set as nodes, road segment connections are set as edges, and an adjacency matrix is ​​constructed based on geographical location relationships and road connectivity relationships. Simultaneously, each node is assigned traffic flow, average vehicle speed, and occupancy characteristics to form a spatiotemporal graph dataset for subsequent modeling. Before constructing the spatiotemporal graph dataset, the original traffic data can be preprocessed, including replacing missing values, generating time-series samples, and normalizing the input data. Specifically, the model input samples are constructed by predicting the next time-series data using preceding time window data. In this embodiment, the spatiotemporal graph dataset is input into the graph convolutional network for spatial feature learning. The neighborhood features of each node are convolved and aggregated through the adjacency matrix, so that the updated feature representation of each node not only includes the traffic state information of the node itself, but also the spatial association information of the neighboring nodes. This allows the spatial correlation of traffic flow between different road areas to be extracted, and a spatial feature sequence representing the spatial dependence of road areas to be obtained. Furthermore, the Long Short-Term Memory (LSTM) network controls the flow of information through input gates, forget gates, and output gates, and remembers and updates historical temporal information through cell states. In an optional implementation, the spatial features output by the graph convolutional network can be used as the input features of each time step. After the LTM network extracts the temporal features, they are then input into the fully connected layer to output the traffic flow prediction results for the target time period, thereby realizing the spatiotemporal prediction of traffic flow. After obtaining the traffic flow prediction results, the traffic flow prediction results and the historical charging demand in the charging station operation data are jointly fitted based on the Copula dependency model to establish a nonlinear correlation between traffic flow and charging demand. Specifically, a correlation dataset can be constructed based on the traffic flow prediction results and historical charging demand. The maximum likelihood estimation method is used to fit the marginal distribution of traffic flow and charging demand respectively. Then, the joint probability distribution of the two is constructed based on the Gaussian Copula function, and the corresponding conditional charging demand distribution is obtained under the given traffic flow conditions to obtain the charging demand for the target time period. After obtaining the distribution of charging demand under certain conditions, the response potential of each charging station is evaluated by combining the station resource status data. The response potential evaluation includes assessing the load capacity and response speed of each charging station under different traffic flow conditions to determine the degree to which each charging station meets the charging demand, and outputting the corresponding station response potential evaluation results. Furthermore, charging resources can be optimized and scheduled based on the station response potential evaluation results. For example, charging stations with loads below the average charging condition can be prioritized during peak traffic periods, and the overload of charging piles can be avoided by dynamically adjusting the working status of charging piles, so as to achieve a balanced distribution of charging load and improve the response capability and resource utilization of charging stations. Through the above implementation methods, this embodiment uses graph convolutional networks and long short-term memory networks to achieve spatiotemporal prediction of traffic flow, and uses the Copula dependency model to establish the correlation between traffic flow and charging demand, thereby completing the assessment of the response potential of electric vehicle charging stations and the optimization of resource scheduling, so as to more effectively reflect the impact of traffic fluctuations on charging demand and station response capabilities.

[0024] The technical solution of this invention constructs a road traffic map and uses graph convolutional networks and long short-term memory networks to predict traffic flow in the time and space. Then, it uses the Copula model to establish the correlation between traffic flow and charging demand, thereby realizing the assessment of the response potential of electric vehicle charging stations and the optimization of resource scheduling. This effectively solves the problem that existing technologies are unable to accurately assess the response potential of electric vehicle charging stations and achieve efficient resource scheduling under the conditions of spatiotemporal fluctuations in traffic flow and nonlinear correlation of charging demand.

[0025] Furthermore, a spatiotemporal graph dataset representing the relationships between road regions is constructed, including: Missing values ​​in road traffic monitoring data are handled and replaced with 0. Sequence samples were generated from the road traffic monitoring data after missing values ​​were processed, and the data for the next moment was predicted using the first 50 time series data. The road traffic monitoring data generated from the sequence samples were normalized to obtain the input data for constructing the spatiotemporal graph dataset.

[0026] As a preferred embodiment of the above, the spatial features of traffic flow are extracted using GCN. First, the road traffic monitoring data is processed as follows: Missing value handling: Replace missing values ​​with 0. This means that the distance information between all sensors is not included, only a small part is included. Replace all unknown distance data with 0. Sequence data generation: The training set and the test set were divided, and the first 50 time series were used to predict the next data. Data normalization: Normalization is performed using MinMaxScaler; ; In the formula, The result is after normalization, where x is the actual result. min and x max These are the minimum and maximum values, respectively.

[0027] Furthermore, constructing a spatiotemporal graph dataset representing the relationships between road regions also includes: The road network is abstracted as a graph structure, with road segments as nodes and the connection relationships between road segments as edges; Construct an adjacency matrix corresponding to the graph structure based on geographical location relationships and road connectivity relationships; Assign nodes current traffic flow, average vehicle speed, and occupancy characteristics to form a spatiotemporal graph dataset.

[0028] As a preferred embodiment of the above embodiments, in one preferred embodiment, there are N=307 sensor nodes based on the data, and each node records three types of traffic data at each time step: flow, occupancy, and speed; therefore, at a certain time step t, the characteristics of all sensor nodes can be represented as a matrix: ; In the formula, This represents the nodal feature matrix at time t. This represents an N-row, F-column matrix space consisting of real numbers, where N=307 represents the number of sensors and F=3 represents the feature dimensions (flow rate, occupancy rate, speed) of each node. To capture the spatial relationships between sensor nodes, a graph G=(V,E) needs to be constructed, where V= E is the set of nodes, and E is the set of edges; each node There may be a feature vector ,in It is typically used to represent a d-dimensional real space; The existence and weight of edges reflect the geographical distance or other correlations between sensors, specifically: Define an adjacency matrix A to represent the structure of a graph, denoted as: If there is an edge between node i and node j, then ,otherwise In practice, since distance information may be missing, missing distance values ​​are replaced with 0 to indicate that there is no direct connection between these sensors. The formula for graph convolution is: ; In the formula, It is the node feature matrix of the Lth layer. 50 represents the time window number, meaning the 51st data is predicted using the first 50 data points, and the next data is predicted by rolling the time window. 3 represents the three features of lane traffic flow, occupancy rate, and vehicle speed, meaning that the spatial dimension extracted after the GCN convolution operation is 25. It is the initial feature matrix of the node; It is a normalized adjacency matrix, where D is the degree matrix. ; It is the learnable weight matrix of the l-th layer. , where 25 represents the GCN convolution channels; σ is a non-linear activation function (e.g., ReLU); Through this operation, the new feature representation of each node not only takes into account its own original features, but also incorporates the feature information of its neighboring nodes; This formula represents the expression obtained through the adjacency matrix. This aggregates the neighbor information of each node and uses a learnable weight matrix. This process transforms features to ultimately generate new features for the nodes. First, traffic flow data is processed using GCN to extract spatial features of traffic flow information from each sensor. If we assume there are three nodes in total, then the nodes The input feature matrix is , = Then node The GCN output is The output of GCN extended to 307 nodes is as follows: ; in The value is 25, which is the dimension of the spatial features extracted by GCN. As input to the LSTM.

[0029] Furthermore, the learned spatial feature sequence includes: Input the spatiotemporal graph dataset into the graph convolutional network; Convolutional aggregation is performed on the features of the neighboring nodes of each node based on the adjacency matrix; Spatial features of nodes that incorporate neighborhood information are generated by learningable weight matrices and nonlinear activation functions to obtain spatial feature sequences.

[0030] As a preferred embodiment of the above, LSTM is used to perform time series modeling on traffic flow data to capture the time dependence of traffic flow. At the core of LSTM is a cell state, which spans the entire time series and is responsible for the transfer of long-term memory. It dynamically regulates the information flow through three gating mechanisms (input gate, forget gate, and output gate). The formula for the weight of the forget gate: ; In the formula, σ is a nonlinear activation function (e.g., ReLU). Here is the forget gate weight matrix. This is the hidden state from the previous step. The input data is at time t. For the offset of the forget gate; The input gate consists of two steps: Step 1: Candidate value generation: Generate candidate memories using the Tanh function, as shown in the following formula: ; In the formula, The candidate memory weight matrix, Bias for candidate memories; The second step is input filtering: use the Sigmoid function to determine the proportion of candidate values ​​to retain, as shown in the following formula: ; In the formula, σ is a nonlinear activation function (e.g., ReLU). The candidate input gate weight matrix, This is the bias of the input gate; Cell state update: The formula for updating the cell state at each time step is: ; In the formula, This represents the cell state at time t. This represents the cell state at time t-1. This represents the output of the forget gate at time t. The input gate output represents the time t. This represents the candidate cell state at time t, where t represents the current time step; The function of the output gate is to determine the final output based on the current cell state and input.

[0031] ; ; In the formula, The candidate memory weight matrix, Bias for candidate memories Let t be the output value of the output gate. It is the hidden state at time t; LSTM uses a forget gate to control the decay of historical information and an input gate to control the addition of new information, in order to further extract the temporal features of traffic flow.

[0032] Furthermore, the traffic flow prediction results for the target time period include: The spatial feature sequence is input into the first layer of the long short-term memory network, and the intermediate states at each time step are retained; The output of the first layer of the Long Short-Term Memory network is input into the second layer of the Long Short-Term Memory network, and the hidden state of the last time step is output. Input the hidden layer data output from the Long Short-Term Memory network into the fully connected layer; Traffic flow prediction results for the target time period are obtained based on the fully connected layer.

[0033] As a preferred embodiment of the above, GCN extracts spatial features and then inputs them into LSTM. After GCN processing, the features of each node at time step t are represented as follows: ; Represents a node The feature vector extracted from spatial features at time step t is used as the input to the LSTM. , R is the feature dimension output by GCN, N represents the number of nodes, and R is a matrix space composed of real numbers.

[0034] LSTM then extracts temporal features. In summary, the formula for GCN-LSTM is as follows: ; ; ; in, , These represent the first and second layers of the LSTM, respectively. The first layer of the LSTM treats each node as a time step and retains all intermediate states; the second layer of the LSTM only outputs the hidden state of the last time step, compressing the temporal dimension and aggregating the global spatiotemporal pattern. It is a normalized adjacency matrix; This represents the temporal features or hidden state of the first-layer LSTM output at time t; This represents the hidden state of the first LSTM layer in the previous time step, which is used to update the output at the current time step together with the current input; This represents the temporal features or hidden state of the second-layer LSTM output at time t; This represents the hidden state of the second LSTM layer in the previous time step, which is used to recursively update the output of the second layer at the current time step. Hidden layer data output after passing through the GCN-LSTM model The final prediction is obtained after passing through a fully connected layer: ; in It is the weight matrix of the fully connected layer. The bias of the fully connected layer; the fully connected layer decouples the 50-dimensional spatiotemporal encoding vector into 307 independent flow estimates through linear transformation and nonlinear activation, where tv is the final predicted flow. Figure 3 As shown in Table 1, the error assessment for predicted traffic flow is as follows: .

[0035] Furthermore, such as Figure 4 As shown, a joint distribution fitting of historical charging demand is performed, including: A correlation dataset is constructed based on the traffic flow prediction results and historical data of charging stations. The correlation dataset includes the traffic flow and corresponding charging demand at each time point. The maximum likelihood estimation method was used to fit the marginal distribution of traffic flow and charging demand, respectively. A joint probability distribution of traffic flow and charging demand is constructed based on the Gaussian Copula function, and the charging demand is predicted given the traffic flow to obtain the conditional charging demand distribution.

[0036] As a preferred embodiment of the above, a dataset is constructed based on the predicted traffic flow data and historical charging station data in the aforementioned scheme. ,in This represents the traffic flow at the i-th time point. This indicates the corresponding charging requirement; First, determine the marginal distribution: Assume that the traffic flow tv follows a normal distribution N( , Assume that the charging demand cd follows a log-normal distribution logN( , ), , and , Given the mean and standard deviation of traffic flow and charging demand, respectively, the standardization transformation is as follows: , ; in, This represents the standardized traffic flow result corresponding to the i-th sample. This represents the standardized charging demand for the i-th sample, where the subscript i indicates the number of samples. The Gaussian Copula is a commonly used Copula type, suitable for modeling linear correlations, and its formula is: ; Where u represents the standardized conversion result of traffic flow, and v represents the standardized conversion result of charging demand. It is the CDF of the standard normal distribution. It is its inverse function. It is the correlation coefficient, which represents the linear correlation between traffic flow and charging demand; For a Gaussian Copula, the joint probability density function is: = ); in , , and , These are the mean and standard deviation of traffic flow and charging demand, respectively. Inverse transformation generates prediction samples: Based on the relationship between traffic flow and charging demand data, charging demand can be predicted given traffic flow.

[0037] ; In the formula, This indicates the charging demand when the traffic flow value is tv0. The inverse cumulative distribution function represents the marginal distribution of charging demand, and the charging demand cd of the charging station is obtained from this function.

[0038] Furthermore, obtain the response potential indicators of each charging station during the target time period, including: The charging demand of charging stations is obtained based on the distribution of charging demand under conditions. Predict the demand response volume of charging stations based on charging demand and station resource status data; The load capacity and response speed of each charging station under different traffic flow conditions are assessed based on the demand response volume to determine the degree to which charging demand is met and to form the assessment results of the station's response potential.

[0039] As a preferred embodiment of the above, after completing the correlation modeling between traffic flow prediction results and charging demand, the charging demand of each charging station during the target time period is obtained according to the conditional charging demand distribution, and the demand response volume is further predicted by combining the site resource status data of each charging station; on this basis, the load capacity and response speed of each charging station under different traffic flow conditions are evaluated to determine the degree to which each charging station meets the charging demand, thereby forming a site response potential evaluation result; through the above method, traffic flow changes, charging demand changes and site operation capacity can be uniformly incorporated into the evaluation process, thereby providing a basis for subsequent resource scheduling.

[0040] Furthermore, the output includes the corresponding site response potential assessment results, including: Identify the peak traffic periods corresponding to the target time period based on the station's response potential assessment results; During peak traffic hours, charging stations with loads below average charging conditions will be prioritized for dispatch. By dynamically adjusting the working status of charging piles, overload can be avoided, thereby optimizing the allocation of charging resources.

[0041] As a preferred embodiment of the above, after obtaining the evaluation results of the station response potential, the load status of each charging station is identified according to the peak traffic period of the target time. Priority is given to scheduling and allocating charging stations with loads below the average charging condition. The overload of charging piles is avoided by dynamically adjusting the working status of the charging piles, thereby achieving optimized allocation of charging resources, balanced distribution of charging load, and improved response capability of charging stations.

[0042] Example 2; Based on the same inventive concept as the charging station response evaluation method based on spatiotemporal flow correlation in the foregoing embodiments, the present invention also provides a charging station response evaluation system based on spatiotemporal flow correlation, the system comprising: The data acquisition module acquires road traffic monitoring data, charging station operation data, and station resource status data, and constructs a spatiotemporal map dataset representing the correlation between road areas based on the road traffic monitoring data; The feature learning module inputs the spatiotemporal graph dataset into the graph convolutional network to learn spatial features and obtain spatial feature sequences that characterize the correlation of traffic flow in different road areas. The traffic flow prediction module inputs the spatial feature sequence into a long short-term memory network for time-dependent modeling to obtain the traffic flow prediction results for the target time period. The distribution fitting module uses the Copula dependency model to perform joint distribution fitting between the traffic flow prediction results and the historical charging demand in the charging station operation data to obtain the conditional charging demand distribution for the target time period. The potential assessment module calculates the response potential index of each charging station during the target time period based on the distribution of conditional charging demand and the status data of station resources, and outputs the corresponding station response potential assessment results.

[0043] The adjustment system described above in this invention can effectively realize the charging station response evaluation method based on spatiotemporal flow correlation, and the technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0044] Example 3; Based on the same inventive concept as the charging station response evaluation method based on spatiotemporal flow correlation in the foregoing embodiments, the present invention also provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, the computer program including program instructions, which, when executed by a processor, can realize the charging station response evaluation method based on spatiotemporal flow correlation.

[0045] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for evaluating the response of charging stations based on spatiotemporal traffic correlation, characterized in that, The method includes: Acquire road traffic monitoring data, charging station operation data, and station resource status data, and construct a spatiotemporal map dataset representing the correlation between road areas based on the road traffic monitoring data; The spatiotemporal graph dataset is input into a graph convolutional network for spatial feature learning to obtain a spatial feature sequence representing the correlation of traffic flow in different road areas; The spatial feature sequence is input into a long short-term memory network for time-dependent modeling to obtain the traffic flow prediction result for the target time period; Based on the Copula dependency model, the traffic flow prediction results are jointly fitted with the historical charging demand in the charging station operation data to obtain the conditional charging demand distribution for the target time period. Based on the conditional charging demand distribution and the site resource status data, the response potential index of each charging site during the target time period is calculated, and the corresponding site response potential assessment result is output.

2. The charging station response evaluation method based on spatiotemporal traffic correlation according to claim 1, characterized in that, Construct a spatiotemporal graph dataset representing the relationships between road regions, including: The missing values ​​in the road traffic monitoring data are processed and replaced with 0. The road traffic monitoring data after missing value processing is used to generate sequence samples, and the data of the next moment is predicted using the first 50 time series data. The road traffic monitoring data generated from the sequence samples is normalized to obtain the input data used to construct the spatiotemporal graph dataset.

3. The charging station response evaluation method based on spatiotemporal traffic correlation according to claim 2, characterized in that, Constructing a spatiotemporal graph dataset representing the relationships between road regions also includes: The road network is abstracted as a graph structure, with road segments as nodes and the connection relationships between road segments as edges; Construct the adjacency matrix corresponding to the graph structure based on geographical location relationships and road connectivity relationships; The nodes are assigned current traffic flow, average vehicle speed, and occupancy characteristics to form the spatiotemporal graph dataset.

4. The charging station response evaluation method based on spatiotemporal traffic correlation according to claim 1, characterized in that, The learned spatial feature sequence includes: The spatiotemporal graph dataset is input into the graph convolutional network; Convolutional aggregation is performed on the features of the neighboring nodes of each node based on the adjacency matrix; The spatial feature sequence is obtained by generating node spatial features that incorporate neighborhood information through a learnable weight matrix and a nonlinear activation function.

5. The charging station response evaluation method based on spatiotemporal traffic correlation according to claim 1, characterized in that, The traffic flow forecast results for the target time period are obtained, including: The spatial feature sequence is input into the first layer of the long short-term memory network, and the intermediate states at each time step are retained; The output of the first layer of the Long Short-Term Memory network is input into the second layer of the Long Short-Term Memory network, and the hidden state of the last time step is output. The hidden layer data output by the Long Short-Term Memory network is input into the fully connected layer; The traffic flow prediction result for the target time period is obtained based on the fully connected layer.

6. The charging station response evaluation method based on spatiotemporal traffic correlation according to claim 1, characterized in that, Joint distribution fitting of historical charging demand includes: A correlation dataset is constructed based on the traffic flow prediction results and historical data of charging stations, wherein the correlation dataset includes traffic flow and corresponding charging demand at each time point; The maximum likelihood estimation method is used to fit the marginal distribution of the traffic flow and the charging demand, respectively. The joint probability distribution of the traffic flow and the charging demand is constructed based on the Gaussian Copula function, and the charging demand is predicted given the traffic flow to obtain the conditional charging demand distribution.

7. The charging station response evaluation method based on spatiotemporal traffic correlation according to claim 1, characterized in that, Obtain response potential indicators for each charging station during the target time period, including: The charging demand of the charging station is obtained based on the charging demand distribution under the aforementioned conditions. Predict the demand response volume of the charging station based on the charging demand and the station resource status data; The load capacity and response speed of each charging station under different traffic flow conditions are evaluated based on the demand response volume to determine the degree to which charging demand is met and to form the evaluation result of the station response potential.

8. The charging station response evaluation method based on spatiotemporal traffic correlation according to claim 7, characterized in that, Output the corresponding site response potential assessment results, including: Based on the assessment results of the station's response potential, identify the peak traffic periods corresponding to the target time period; According to the aforementioned peak traffic periods, charging stations with loads below average charging conditions will be prioritized for scheduling. By dynamically adjusting the working status of charging piles, overload can be avoided, thereby optimizing the allocation of charging resources.

9. A charging station response evaluation system based on spatiotemporal traffic correlation, characterized in that, The system includes: The data acquisition module acquires road traffic monitoring data, charging station operation data, and station resource status data, and constructs a spatiotemporal map dataset representing the correlation between road areas based on the road traffic monitoring data; The feature learning module inputs the spatiotemporal graph dataset into the graph convolutional network to learn spatial features and obtain spatial feature sequences that characterize the correlation of traffic flow in different road areas. The traffic flow prediction module inputs the spatial feature sequence into a long short-term memory network for time-dependent modeling to obtain the traffic flow prediction results for the target time period. The distribution fitting module uses the Copula dependency model to perform joint distribution fitting between the traffic flow prediction results and the historical charging demand in the charging station operation data to obtain the conditional charging demand distribution for the target time period. The potential assessment module calculates the response potential index of each charging station during the target time period based on the distribution of conditional charging demand and the status data of station resources, and outputs the corresponding station response potential assessment results.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which includes program instructions that, when executed by a processor, can implement the charging station response evaluation method based on spatiotemporal traffic correlation as described in any one of claims 1-8.

Images