Parking recommendation method and system of electric vehicle and related equipment

Abstract

The invention provides a parking recommendation method and system for an electric vehicle and related equipment. The method comprises the steps of collecting historical riding data of the electric vehicle of a user; training a parking prediction hybrid model through the historical riding data, wherein the parking prediction hybrid model comprises a recurrent neural network and a space-time diagram neural network; predicting each riding of the user through the parking prediction hybrid model; when it is predicted that the user needs to park the vehicle, generating a vehicle taking prompt and obtaining surrounding parking point information; scoring and sorting the surrounding parking point information by using a preset parking point recommendation algorithm to obtain a parking recommendation list; and displaying the parking recommendation list for a user to select. According to the technical scheme, the trained model is constructed to perform parking prediction reminding on the user and provide the recommended parking point, so that the user can be prevented from missing a proper parking position during riding, and the riding experience of the user is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of parking management, and in particular to a parking recommendation method, system and related equipment for electric vehicles. Background Art

[0002] With the increasing popularity of electric vehicles, parking management is facing significant challenges, such as limited parking space, low parking efficiency, and safety hazards caused by irregular parking. Currently, electric vehicle parking is primarily based on human intervention and visual recognition to determine whether a parking spot is available. Existing parking methods for electric vehicles also suffer from a lack of planning and inefficiency. Therefore, existing technologies are developing a method that can predict user parking locations and guide users to park their vehicles.

[0003] Therefore, the existing technology still needs to be improved and developed. Summary of the Invention

[0004] The present invention provides a parking recommendation method, system and related equipment for electric vehicles. The main purpose of the present invention is to solve the technical problems mentioned in the background technology of the existing technology.

[0005] A first aspect of the present invention provides a parking recommendation method for an electric vehicle, comprising:

[0006] Collect historical riding data of the user's electric vehicle, including trajectory data, driving status data, environmental data, and user profile data;

[0007] Training a parking prediction hybrid model using the historical riding data, wherein the parking prediction hybrid model includes a recurrent neural network and a spatiotemporal graph neural network;

[0008] Predicting each ride of the user using the parking prediction hybrid model;

[0009] When it is predicted that the user needs to park, a car pick-up reminder is generated and surrounding parking spot information is obtained;

[0010] Using a preset parking spot recommendation algorithm to score and sort the surrounding parking spot information to obtain a parking recommendation list;

[0011] The parking recommendation list is displayed for the user to select.

[0012] In an optional implementation of the first aspect of the present invention, the training of the parking prediction hybrid model using the historical riding data includes:

[0013] Based on the historical riding data, the recurrent neural network is trained to capture the long-term dependency of the time series to obtain a periodic pattern for predicting the user's parking time point;

[0014] Based on the historical riding data, the spatiotemporal graph neural network is subjected to synchronous spatial modeling combining graph structure and time series to obtain predicted user parking time dynamics and modeled spatial proximity.

[0015] In an optional implementation of the first aspect of the present invention, the training of the recurrent neural network based on the historical riding data to capture the long-term dependency of the time series to obtain a periodic pattern for predicting the user's parking time point includes:

[0016] Performing temporal behavior feature extraction, environmental feature extraction, spatial feature extraction, and user portrait feature extraction on the historical riding data;

[0017] Performing a first preprocessing on the temporal behavior feature, the environmental feature, the spatial feature, and the user portrait feature, wherein the first preprocessing includes temporal normalization, feature decomposition, and feature encoding;

[0018] Extracting periodicity, trend, and random fluctuation components from the temporal behavior features, the environmental features, the spatial features, and the user portrait features;

[0019] Based on the extracted periodicity, trend and random fluctuation components, the long-term dependency of the user's parking behavior is captured by using a long short-term memory unit.

[0020] In an optional embodiment of the first aspect of the present invention, performing synchronous spatial modeling combining graph structure and time series on the spatiotemporal graph neural network based on the historical riding data to obtain predicted user parking time dynamics and model spatial proximity includes:

[0021] Extracting temporal behavior features, spatial topology features, environmental features, and user interaction features from the historical riding data;

[0022] performing a second preprocessing on the temporal behavior feature, the spatial topology feature, the environmental feature, and the user interaction feature, wherein the second preprocessing includes spatiotemporal coding, dynamic graph construction, and multimodal fusion;

[0023] A hypergraph convolution unit is used to process the user-time-place-behavior interaction relationship among the temporal behavior features, the spatial topology features, the environmental features, and the user interaction features to construct a hyperedge multimodal association model.

[0024] In an optional embodiment of the first aspect of the present invention, the use of a hypergraph convolution unit to process the user-time-place-behavior interaction relationship between the temporal behavior features, the spatial topology features, the environmental features, and the user interaction features to construct a hyperedge multimodal association model includes:

[0025] Based on the temporal behavior characteristics, the spatial topology characteristics, the environmental characteristics and the user interaction characteristics, a hyperedge is defined, and an association matrix, a weight matrix, a vertex degree matrix and a hyperedge degree matrix are constructed;

[0026] Obtaining a Laplacian matrix based on the incidence matrix, the weight matrix, the vertex degree matrix, and the hyperedge degree matrix;

[0027] A hypergraph convolution operation is performed on the Laplacian matrix to obtain a simplified graph adjacency matrix.

[0028] In an optional implementation of the first aspect of the present invention, the scoring and sorting of the surrounding parking spot information using a preset parking spot recommendation algorithm to obtain a parking recommendation list includes:

[0029] Based on the surrounding parking spot information, calculate the distance between each surrounding parking spot and the current position of the electric vehicle and sort them from small to large;

[0030] Selecting a number of pre-screened parking spots ranked higher in distance from the surrounding parking spots;

[0031] Evaluate the parking congestion level of the plurality of pre-screened parking spots and sort them from least crowded to most crowded;

[0032] Selecting a number of secondary screening parking spots ranked higher in terms of congestion from the number of primary screening parking spots;

[0033] Traffic route flow is evaluated for several of the screened parking spots and sorted from small to large to obtain a parking recommendation list.

[0034] In an optional implementation of the first aspect of the present invention, after displaying the parking recommendation list for user selection, the method further includes:

[0035] receiving a user's selection of a target parking spot in the parking recommendation list;

[0036] Acquiring location data of the target parking point;

[0037] The location data is input into a preset navigation software to generate a path plan to the target parking point.

[0038] A second aspect of the present invention provides a parking recommendation system for an electric vehicle, the parking recommendation system for an electric vehicle comprising:

[0039] A data collection module is used to collect historical riding data of the user's electric vehicle, wherein the historical riding data includes trajectory data, driving status data, environmental data and user portrait data;

[0040] a model training module, configured to train a parking prediction hybrid model using the historical riding data, wherein the parking prediction hybrid model includes a recurrent neural network and a spatiotemporal graph neural network;

[0041] A parking prediction module, configured to predict each ride of the user using the parking prediction hybrid model;

[0042] The parking reminder module is used to generate a car pick-up reminder and obtain surrounding parking spot information when it is predicted that the user needs to park;

[0043] A parking recommendation module is used to score and sort the surrounding parking spot information using a preset parking spot recommendation algorithm to obtain a parking recommendation list;

[0044] The recommendation display module is used to display the parking recommendation list for the user to select.

[0045] A third aspect of the present invention provides a central control device, comprising: a memory and at least one processor, wherein the memory stores instructions, and the memory and the at least one processor are interconnected via a line;

[0046] The at least one processor calls the instructions in the memory to enable the central control device to execute the parking recommendation method for an electric vehicle as described in any one of the first aspects of the present invention.

[0047] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the parking recommendation method for an electric vehicle as described in any one of the first aspects of the present invention is implemented.

[0048] Beneficial effects: The present invention provides a parking recommendation method, system, and related equipment for electric vehicles. The method includes collecting historical riding data of a user's electric vehicle; training a parking prediction hybrid model using the historical riding data, wherein the parking prediction hybrid model includes a recurrent neural network and a spatiotemporal graph neural network; predicting each ride of the user using the parking prediction hybrid model; generating a vehicle pick-up reminder and obtaining surrounding parking spot information when it is predicted that the user needs to park; scoring and sorting the surrounding parking spot information using a preset parking spot recommendation algorithm to obtain a parking recommendation list; and displaying the parking recommendation list for the user to select. The technical solution of the present invention prevents users from missing suitable parking locations by constructing a trained model to provide parking prediction reminders and recommended parking spots, thereby improving the user's riding experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic diagram of an embodiment of the main method steps of a parking recommendation method for an electric vehicle according to the present invention;

[0050] Figure 2 A schematic diagram of an embodiment of a parking recommendation system for electric vehicles according to the present invention;

[0051] Figure 3 The figure is a schematic diagram of an embodiment of a central control device of the present invention. DETAILED DESCRIPTION

[0052] The terms "first," "second," "third," "fourth," and the like (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatus.

[0053] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 In a first aspect, the present invention provides a parking recommendation method for an electric vehicle. The method may be executed by a mobile terminal (e.g., a mobile phone) and a central control device. The method includes:

[0054] S100. Collect historical riding data of the user's electric vehicle, wherein the historical riding data includes trajectory data, driving status data, environmental data, and user portrait data. In the present invention, the trajectory data includes each historical riding trajectory of the user recorded by the system, and the driving status data includes information such as the start and stop, speed, and acceleration of the vehicle at each time point of each riding trajectory. The environmental data includes data such as weather conditions, traffic environment, and surrounding activities during each riding trajectory, and the user portrait data includes data such as occupation, residence / work location, and preferences.

[0055] S200. Training a parking prediction hybrid model using the historical riding data, the parking prediction hybrid model including a recurrent neural network and a spatiotemporal graph neural network. In the present invention, in order for the parking prediction hybrid model to better predict user parking time, it is necessary to train the model using the collected trajectory data, driving status data, environmental data, and user profile data, thereby obtaining a prediction model that can capture spatiotemporal dependencies and user habits. The parking prediction hybrid model of the present invention combines a recurrent neural network and a spatiotemporal graph neural network. The recurrent neural network excels at capturing long-term dependencies in time series and is suitable for predicting the periodic laws of user parking time points. The spatiotemporal graph neural network can combine graph structures (such as parking lot nodes and roadside edges) and time series to simultaneously model spatial proximity (such as the association between parking lots frequently visited by users) and temporal dynamics. In this way, the parking prediction hybrid model of the present invention can predict user parking under more complex conditions, can use parking estimates under different scenario conditions, and obtain more accurate prediction results.

[0056] In an optional implementation of step S200 of the present invention, the training of the parking prediction hybrid model using the historical riding data includes:

[0057] S201. Based on the historical riding data, the recurrent neural network is trained to capture the long-term dependency of the time series, so as to obtain the periodic law for predicting the user's parking time point; specifically, this step may include: extracting temporal behavior features, environmental features, spatial features and user portrait features from the historical riding data; performing a first preprocessing on the temporal behavior features, the environmental features, the spatial features and the user portrait features, the first preprocessing including temporal normalization, feature decomposition and feature encoding; extracting periodicity, trend and random fluctuation components from the temporal behavior features, the environmental features, the spatial features and the user portrait features; based on the extracted periodicity, trend and random fluctuation components and using long short-term memory units, the long-term dependency of the user's parking behavior is captured.

[0058] In the present invention, the temporal behavioral features mainly refer to serial data such as the user's historical parking time, parking location distribution, parking duration and parking frequency; the environmental features mainly refer to time-related variables such as weather (such as rainfall, temperature), holidays, weekdays / weekends, and surrounding activities (such as concerts); the spatial features mainly refer to the location of the parking lot, parking space capacity, and congestion of adjacent roads, which can be expressed through geocoding or graph structure; the user portrait mainly refers to the construction of multidimensional features by combining static attributes (such as occupation, residence) and dynamic behaviors (such as high-frequency parking periods). The main purpose of the first feature preprocessing is to normalize the time series, standardize the format and vectorize the features to facilitate subsequent processing by the neural network. After the feature preprocessing, the periodicity (such as morning and evening peak hours), trend (such as increased parking demand on weekends) and random fluctuation components of the features are extracted as specific analysis factors. Finally, the long short-term memory unit (LSTM unit) is used to capture the long-term dependencies of parking behaviors (such as commuting parking patterns for consecutive days). During the training process, back propagation along time can be used to solve the problem of gradient disappearance, regularization technology can be used to solve problems in data sparse scenarios, and transfer learning methods can be used to make model parameters converge quickly. Through the training in the above manner, the parking prediction hybrid model of the present invention has the ability to predict user parking in time series scenarios.

[0059] S202. Based on the historical riding data, the spatiotemporal graph neural network is subjected to synchronous spatial modeling combining graph structure and time series to obtain the predicted user parking time dynamics and modeling spatial proximity. Specifically, this step may include: extracting temporal behavior features, spatial topology features, environmental features, and user interaction features from the historical riding data; performing a second preprocessing on the temporal behavior features, the spatial topology features, the environmental features, and the user interaction features, the second preprocessing including spatiotemporal coding, dynamic graph construction, and multimodal fusion; using a hypergraph convolution unit to process the user-time-place-behavior interaction relationship between the temporal behavior features, the spatial topology features, the environmental features, and the user interaction features to construct a hyperedge multimodal association model.

[0060] In the present invention, the temporal behavior features also include serial data such as the user's historical parking time, parking location distribution, parking duration and parking frequency. The spatial topological features mainly include graph structure data composed of road connections, geographical distances and parking space capacity between parking nodes; the environmental features also mainly include weather, holidays, real-time traffic flow and sudden activities; the user interaction features include user-location and user-time interaction matrices to capture user preferences and habits; the purpose of feature spatiotemporal encoding in the second preprocessing is mainly to divide time into periodic slots and discretize space into regional grids or cluster nodes; the main purpose of dynamic graph construction is to adjust the adjacency matrix weights according to real-time traffic status (such as the road network congestion index affects the node connection strength); the main purpose of multimodal fusion is to fuse other feature data with user portraits and unify them through the embedding layer.

[0061] In the present invention, the use of a hypergraph convolution unit to process the user-time-place-behavior interaction relationship between the temporal behavior characteristics, the spatial topology characteristics, the environmental characteristics, and the user interaction characteristics to construct a hyperedge multimodal association model can specifically include: defining hyperedges based on the temporal behavior characteristics, the spatial topology characteristics, the environmental characteristics, and the user interaction characteristics, and constructing an association matrix, a weight matrix, a vertex degree matrix, and a hyperedge degree matrix; obtaining a Laplacian matrix based on the association matrix, the weight matrix, the vertex degree matrix, and the hyperedge degree matrix; and performing a hypergraph convolution operation on the Laplacian matrix to obtain a simplified graph adjacency matrix. Specifically, the core of hypergraph convolution is to convert the hypergraph into a simple graph structure. When constructing the hypergraph, hyperedges are defined according to data features (such as taking data groups that participate in the same event as hyperedges), and vertex sets and hyperedge sets are obtained. The association matrix, weight matrix, vertex degree matrix and hyperedge degree matrix are constructed based on the vertex set and hyperedge set. The association matrix represents the affiliation between nodes and hyperedges, the weight matrix represents the importance of hyperedges, and the vertex degree matrix is a diagonal matrix that reflects that the diagonal elements of the matrix are the degrees of each vertex: the hyperedge degree matrix is also a diagonal matrix, but it reflects that the diagonal elements of the matrix are the degrees of each hyperedge; in the entire hypergraph convolution process of the present invention, the hyperedge simplification process selects the two nodes with the largest feature difference in the hyperedge as representatives to form the edges of a weighted simple graph. The normalized Laplace matrix of the hypergraph is constructed mainly by capturing the nonlinear relationship between nodes through spectral domain convolution, and finally the convolution kernel is used to aggregate neighborhood information on the simplified graph to update the node representation.

[0062] S300. Predict each ride of the user through the parking prediction hybrid model. In the present invention, after the model is trained, the system will obtain the user's location, vehicle state change data, and environmental data during each ride in real time as model input. The parking prediction hybrid model will predict the input data. When the input data meets the matching conditions for the prediction, the parking prediction hybrid model will output a parking prediction result.

[0063] S400. When it is predicted that the user needs to park, a vehicle pick-up reminder is generated and surrounding parking spot information is obtained. In the present invention, when the model predicts that the user needs to park, a parking reminder will be issued. For example, if there is a predicted parking spot ahead, please pay attention to park in time so that the designated parking spot can be selected nearby.

[0064] S500. Use a preset parking spot recommendation algorithm to score and sort the surrounding parking spot information to obtain a parking recommendation list. In the present invention, if the model predicts that the user has parking behavior, the model will call the preset algorithm to recommend a parking location suitable for the user. The recommendation principle is mainly based on the distance from the user, the congestion level of the parking location, and the traffic conditions when driving to the specified parking spot.

[0065] In an optional implementation of step S500 of the present invention, the use of a preset parking point recommendation algorithm to score and sort the surrounding parking point information to obtain a parking recommendation list includes: based on the surrounding parking point information, calculating the distance between each surrounding parking point and the current position of the electric vehicle and sorting them from small to large; selecting a number of preliminary screening parking points with high distance rankings from each of the surrounding parking points; evaluating the parking congestion level of the several preliminary screening parking points and sorting them from small to large; selecting a number of medium screening parking points with high congestion rankings from the several preliminary screening parking points; evaluating the traffic route flow of the several medium screening parking points and sorting them from small to large to obtain a parking recommendation list. In general, the parking spot recommendation algorithm of the present invention calculates the distance from the user's current location to each candidate location, selects the closest locations, and then evaluates the degree of congestion: based on real-time and historical data, evaluates the parking congestion of each candidate location, and finally combines real-time traffic flow: obtains real-time traffic flow information through traffic monitoring data or map service API, and finally scores each candidate location based on factors such as distance and congestion, and selects a group of locations with higher scores for recommendation. It should be noted that the parking recommendation list of the present invention will change based on changes in real-time data until the user selects the target parking location.

[0066] S600: Display the parking recommendation list for user selection. In the present invention, the terminal system's main interface is configured with a vehicle parameter display block, a map block, and a parking list recommendation block. When no parking is predicted, the parking list recommendation block is blank and minimized. After the parking recommendation list is obtained, the parking list recommendation block is adaptively scaled to a preset size, and the parking recommendation list is placed within the parking list recommendation block for user selection.

[0067] In an optional embodiment of the first aspect of the present invention, displaying the recommended parking list for user selection includes: receiving a user selection of a target parking spot from the recommended parking list; obtaining location data for the target parking spot; and inputting the location data into pre-installed navigation software to generate a route plan to the target parking spot. In this invention, once the user has selected a specific parking spot, the system automatically uses an onboard map to generate a driving route to the target parking spot, guiding the user to the target parking spot for parking. In this invention, after the user has parked, the system updates the parking spot's status, availability, and congestion level. Availability of a parking spot refers to whether the spot is available for parking. It should be noted that congestion level of a parking spot, as used in this invention, refers to the ratio of the number of vehicles parked at the spot to the number of vehicles the spot can accommodate. If there are a large number of vacant parking spaces, the congestion level of the spot is low. If the user parks at the spot, the congestion level increases; if the user leaves, the congestion level decreases.

[0068] See also Figure 2 A second aspect of the present invention provides a parking recommendation system for an electric vehicle, the parking recommendation system for an electric vehicle comprising:

[0069] A data collection module 10 is used to collect historical riding data of a user's electric vehicle, wherein the historical riding data includes trajectory data, driving status data, environmental data, and user portrait data;

[0070] a model training module 20 for training a parking prediction hybrid model using the historical riding data, wherein the parking prediction hybrid model includes a recurrent neural network and a spatiotemporal graph neural network;

[0071] a parking prediction module 30, configured to predict each ride of the user using the parking prediction hybrid model;

[0072] The parking reminder module 40 is used to generate a parking reminder and obtain surrounding parking spot information when it is predicted that the user needs to park;

[0073] The parking recommendation module 50 is used to score and sort the surrounding parking spot information using a preset parking spot recommendation algorithm to obtain a parking recommendation list;

[0074] The recommendation display module 60 is used to display the parking recommendation list for the user to select.

[0075] In an optional implementation of the second aspect of the present invention, the model training module includes:

[0076] a recurrent neural network training unit, configured to train the recurrent neural network to capture long-term dependencies of time series based on the historical riding data, so as to obtain a periodic pattern for predicting user parking times;

[0077] A spatiotemporal graph neural network training unit is used to perform synchronous spatial modeling of the spatiotemporal graph neural network combining graph structure and time series based on the historical riding data, so as to obtain the predicted user parking time dynamics and model spatial proximity.

[0078] In an optional implementation of the second aspect of the present invention, the recurrent neural network training unit includes:

[0079] A first feature extraction subunit is configured to extract temporal behavior features, environmental features, spatial features, and user portrait features from the historical riding data;

[0080] a first preprocessing subunit, configured to perform a first preprocessing on the temporal behavior feature, the environmental feature, the spatial feature, and the user portrait feature, wherein the first preprocessing includes temporal normalization, feature decomposition, and feature encoding;

[0081] An information extraction subunit, configured to extract periodicity, trend, and random fluctuation components from the temporal behavior features, the environmental features, the spatial features, and the user portrait features;

[0082] The relationship capture subunit is used to capture the long-term dependency of the user's parking behavior based on the extracted periodicity, trend and random fluctuation components and using a long short-term memory unit.

[0083] In an optional implementation of the second aspect of the present invention, the spatiotemporal graph neural network training unit includes:

[0084] A second feature extraction subunit is used to extract temporal behavior features, spatial topology features, environmental features, and user interaction features from the historical riding data;

[0085] a second preprocessing subunit, configured to perform a second preprocessing on the temporal behavior feature, the spatial topology feature, the environmental feature, and the user interaction feature, wherein the second preprocessing includes spatiotemporal coding, dynamic graph construction, and multimodal fusion;

[0086] The model construction subunit is used to use the hypergraph convolution unit to process the user-time-place-behavior interaction relationship between the temporal behavior characteristics, the spatial topology characteristics, the environmental characteristics and the user interaction characteristics to construct a hyperedge multimodal association model.

[0087] In an optional embodiment of the second aspect of the present invention, the model building subunit includes:

[0088] A hyperedge correlation matrix construction component is used to define hyperedges based on the temporal behavior characteristics, the spatial topology characteristics, the environmental characteristics, and the user interaction characteristics, and to construct an association matrix, a weight matrix, a vertex degree matrix, and a hyperedge degree matrix;

[0089] A Laplacian matrix construction component, configured to obtain a Laplacian matrix based on the incidence matrix, the weight matrix, the vertex degree matrix, and the hyperedge degree matrix;

[0090] The hypergraph convolution operation component is used to perform a hypergraph convolution operation on the Laplacian matrix to obtain a simplified graph adjacency matrix.

[0091] In an optional implementation of the second aspect of the present invention, the parking recommendation module includes:

[0092] A distance calculation and sorting unit, configured to calculate the distance between each of the surrounding parking spots and the current position of the electric vehicle based on the surrounding parking spot information and sort the distances from small to large;

[0093] A parking spot preliminary screening unit, configured to select a number of preliminary screening parking spots ranked top in distance from the surrounding parking spots;

[0094] a congestion evaluation and ranking unit, configured to evaluate the degree of parking congestion of the plurality of pre-screened parking spots and rank them from least to greatest;

[0095] A parking point screening unit is used to select a number of secondary screening parking points ranked high in terms of congestion from the number of primary screening parking points;

[0096] The recommendation list generating unit is used to evaluate the traffic route flow of the plurality of the selected parking spots and sort them from small to large to obtain a parking recommendation list.

[0097] In an optional implementation of the second aspect of the present invention, the parking recommendation system for electric vehicles further includes:

[0098] a selection receiving module for receiving a user's selection of a target parking spot in the parking recommendation list;

[0099] A parking location acquisition module, configured to acquire location data of the target parking spot;

[0100] The parking path guidance module is used to input the position data into a preset navigation software to generate a path plan for the target parking point.

[0101] Figure 3 It is a structural diagram of a central control device provided by an embodiment of the present invention. The central control device may have relatively large differences due to different configurations or performances, and may include one or more processors 70 (central processing units, CPU) (for example, one or more processors) and memories 80, and one or more storage media 90 for storing applications or data (for example, one or more massive storage devices). Among them, the memories and storage media can be temporary storage or permanent storage. The program stored in the storage medium may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the central control device. Furthermore, the processor can be configured to communicate with the storage medium and execute a series of instruction operations in the storage medium on the central control device.

[0102] The central control device of the present invention may further include one or more power supplies 100, one or more wired or wireless network interfaces 110, one or more input and output interfaces 120, and / or one or more operating systems, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be understood by those skilled in the art that Figure 3 The structure of the central control device shown does not constitute a limitation on the central control device, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0103] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the electric vehicle parking recommendation method.

[0104] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system or system or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0105] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc., various media that can store program code.

[0106] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A parking recommendation method for an electric vehicle, characterized in that: include: Collect historical riding data of the user's electric vehicle, including trajectory data, driving status data, environmental data, and user profile data; Training a parking prediction hybrid model using the historical riding data, wherein the parking prediction hybrid model includes a recurrent neural network and a spatiotemporal graph neural network; Predicting each ride of the user using the parking prediction hybrid model; When it is predicted that the user needs to park, a car pick-up reminder is generated and surrounding parking spot information is obtained; Using a preset parking spot recommendation algorithm to score and sort the surrounding parking spot information to obtain a parking recommendation list; The parking recommendation list is displayed for the user to select.

2. The parking recommendation method for electric vehicles according to claim 1, characterized in that: The training of the parking prediction hybrid model using the historical riding data includes: Based on the historical riding data, the recurrent neural network is trained to capture the long-term dependency of the time series to obtain a periodic pattern for predicting the user's parking time point; Based on the historical riding data, the spatiotemporal graph neural network is subjected to synchronous spatial modeling combining graph structure and time series to obtain predicted user parking time dynamics and modeled spatial proximity.

3. The parking recommendation method for electric vehicles according to claim 2, characterized in that: The training of the recurrent neural network based on the historical riding data to capture the long-term dependency of the time series to obtain a periodic law for predicting the user's parking time point includes: Performing temporal behavior feature extraction, environmental feature extraction, spatial feature extraction, and user portrait feature extraction on the historical riding data; Performing a first preprocessing on the temporal behavior feature, the environmental feature, the spatial feature, and the user portrait feature, wherein the first preprocessing includes temporal normalization, feature decomposition, and feature encoding; Extracting periodicity, trend, and random fluctuation components from the temporal behavior features, the environmental features, the spatial features, and the user portrait features; Based on the extracted periodicity, trend and random fluctuation components, the long-term dependency of the user's parking behavior is captured by using a long short-term memory unit.

4. The parking recommendation method for electric vehicles according to claim 2, characterized in that: The performing synchronous spatial modeling combining graph structure and time series on the spatiotemporal graph neural network based on the historical riding data to obtain predicted user parking time dynamics and model spatial proximity includes: Extracting temporal behavior features, spatial topology features, environmental features, and user interaction features from the historical riding data; performing a second preprocessing on the temporal behavior feature, the spatial topology feature, the environmental feature, and the user interaction feature, wherein the second preprocessing includes spatiotemporal coding, dynamic graph construction, and multimodal fusion; A hypergraph convolution unit is used to process the user-time-place-behavior interaction relationship among the temporal behavior features, the spatial topology features, the environmental features, and the user interaction features to construct a hyperedge multimodal association model.

5. The parking recommendation method for electric vehicles according to claim 4, characterized in that: The method of using a hypergraph convolution unit to process the user-time-place-behavior interaction relationship among the temporal behavior feature, the spatial topology feature, the environmental feature, and the user interaction feature to construct a hyperedge multimodal association model includes: Based on the temporal behavior characteristics, the spatial topology characteristics, the environmental characteristics and the user interaction characteristics, a hyperedge is defined, and an association matrix, a weight matrix, a vertex degree matrix and a hyperedge degree matrix are constructed; Obtaining a Laplacian matrix based on the incidence matrix, the weight matrix, the vertex degree matrix, and the hyperedge degree matrix; A hypergraph convolution operation is performed on the Laplacian matrix to obtain a simplified graph adjacency matrix.

6. The parking recommendation method for electric vehicles according to claim 1, characterized in that: The step of scoring and sorting the surrounding parking spot information using a preset parking spot recommendation algorithm to obtain a parking recommendation list includes: Based on the surrounding parking spot information, calculate the distance between each surrounding parking spot and the current position of the electric vehicle and sort them from small to large; Selecting a number of pre-screened parking spots ranked higher in distance from the surrounding parking spots; Evaluate the parking congestion level of the plurality of pre-screened parking spots and sort them from least crowded to most crowded; Selecting a number of secondary screening parking spots ranked higher in terms of congestion from the number of primary screening parking spots; Traffic route flow is evaluated for several of the screened parking spots and sorted from small to large to obtain a parking recommendation list.

7. The parking recommendation method for an electric vehicle according to claim 1, characterized in that: After displaying the parking recommendation list for the user to select, the method further comprises: receiving a user's selection of a target parking spot in the parking recommendation list; Acquiring location data of the target parking point; The location data is input into a preset navigation software to generate a path plan to the target parking point.

8. A parking recommendation system for electric vehicles, characterized in that: The parking recommendation system for electric vehicles includes: A data collection module is used to collect historical riding data of the user's electric vehicle, wherein the historical riding data includes trajectory data, driving status data, environmental data and user portrait data; a model training module, configured to train a parking prediction hybrid model using the historical riding data, wherein the parking prediction hybrid model includes a recurrent neural network and a spatiotemporal graph neural network; A parking prediction module, configured to predict each ride of the user using the parking prediction hybrid model; The parking reminder module is used to generate a car pick-up reminder and obtain surrounding parking spot information when it is predicted that the user needs to park; A parking recommendation module is used to score and sort the surrounding parking spot information using a preset parking spot recommendation algorithm to obtain a parking recommendation list; The recommendation display module is used to display the parking recommendation list for the user to select.

9. A central control device, characterized in that: The central control device includes: a memory and at least one processor, wherein the memory stores instructions, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instruction in the memory to enable the central control device to execute the parking recommendation method for an electric vehicle according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the parking recommendation method for an electric vehicle according to any one of claims 1 to 7 is implemented.

Images