Vehicle resource prediction method and system, electronic equipment and computer program product
By constructing heterogeneous networks and generating vehicle-use relationship characteristics, and combining deep learning models to predict, the problem of low accuracy in vehicle factors in vehicle business is solved, and a more comprehensive vehicle resource prediction is achieved.
Patent Information
- Application Number
- CN202411742750.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-02-25
AI Technical Summary
In vehicle business, the prediction accuracy of vehicle resource prediction is not high because each factor (such as vehicle and user) is cooperative, and the correlation between multiple factors needs to be considered.
By constructing a heterogeneous network, a feature vector of vehicle-use relationship characteristics is generated, a node-related relationship is characterized, and it is input with vehicle-use requirements into a deep learning model for prediction, so as to obtain available vehicle resources at future moments.
The accuracy of vehicle resource prediction is improved, and the impact of vehicle use mechanisms, passengers and drivers on vehicle resource consumption is taken into account, making the prediction more comprehensive.
Smart Images
Figure CN119514804A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of vehicle business processing, and in particular to a vehicle resource prediction method, system, electronic device and computer program product. Background Art
[0002] With the expansion of vehicle business, the dispatch and management of official vehicles have gradually become more perfect, and the management of official vehicles is mainly achieved through intelligent logistics systems, which can include warehousing maintenance of vehicles and drivers, application submission and approval functions of departure and destination, thereby generating a huge amount of vehicle application and usage data.
[0003] At present, vehicle resources are often predicted in advance based on vehicle factors such as vehicle travel time and vehicle status, so that vehicles can be dispatched in advance to meet the increasingly diverse vehicle demand. However, in the vehicle business, various factors (such as vehicles and users) are coordinated. Therefore, the prediction accuracy of vehicle resources based only on vehicle factors is not high. Summary of the invention
[0004] In view of the above problems, the present application provides a vehicle resource prediction method, system, electronic device and computer program product to achieve the purpose of improving the accuracy of vehicle resource prediction. The specific scheme is as follows:
[0005] A first aspect of the present application provides a vehicle resource prediction method, the vehicle resource prediction method comprising:
[0006] Receive multiple vehicle use demands input by a user end and obtain current vehicle use relationship features, wherein the vehicle use relationship features are generated based on the topological structure of a heterogeneous network, wherein the heterogeneous network includes four types of nodes, namely, vehicle use organizations, passengers, license plates, and drivers, wherein different types of nodes have edges, and passenger nodes have edges, wherein the weight of the edges is the number of times the edge relationship between nodes appears in historical vehicle use data, and the vehicle use relationship features are feature vectors that characterize the node association relationship in the heterogeneous network;
[0007] The vehicle usage relationship characteristics and the multiple vehicle usage demands are input into a preset deep learning model for prediction to obtain available vehicle resources at a future moment, wherein the preset deep learning model is trained by the historical vehicle usage data, and the future moment is the next moment of the current moment.
[0008] In a possible implementation, the historical vehicle usage data includes a plurality of different vehicle dispatching tasks, and the heterogeneous network construction process includes:
[0009] Determine the factors that affect vehicle resources from multiple different vehicle dispatching tasks, and determine the correlation between factors of the same type and between factors of different types;
[0010] Determine the node type according to different types of factors, and determine the connection relationship between the nodes according to the association relationship between factors of the same type and the association relationship between factors of different types, wherein the node types include vehicle-using organizations, passengers, license plate numbers, and drivers;
[0011] The topological structure of the heterogeneous network is constructed according to the node types and the connection relationships between the nodes. The topological structure has four types of nodes, including vehicle-using organizations, passengers, license plates and drivers. There are edges between different types of nodes, and there are edges between passenger nodes. The weights of the edges are set to: the number of occurrences of the association relationships corresponding to the node edges in multiple different vehicle dispatching tasks.
[0012] In a possible implementation, the heterogeneous network includes a plurality of vehicle-using organization nodes, a plurality of passenger nodes, a plurality of license plate number nodes, and a plurality of driver nodes.
[0013] There are homogeneous edges between the passenger nodes, there are heterogeneous edges between the vehicle-using organization node and the passenger node, there are heterogeneous edges with a many-to-many relationship between the passenger node and the license plate number node, and there are heterogeneous edges with a many-to-many relationship between the license plate number node and the driver node.
[0014] In a possible implementation, the process of generating the vehicle usage relationship feature is as follows:
[0015] In the heterogeneous network, a plurality of node sequences are generated by randomly walking nodes;
[0016] Filtering at least one target node sequence from the multiple node sequences according to the edge weights in the node sequences;
[0017] After unifying the dimensions of the nodes in the target node sequence, a feature vector corresponding to the target node sequence is extracted, and the feature vector corresponding to the target node sequence is used as the vehicle use relationship feature.
[0018] A second aspect of the present application provides a vehicle resource prediction system, the vehicle resource prediction system comprising:
[0019] A receiving unit, used to receive multiple vehicle use demands input by a user terminal and obtain current vehicle use relationship characteristics, wherein the vehicle use relationship characteristics are generated based on the topological structure of a heterogeneous network, wherein the heterogeneous network includes four types of nodes, namely, vehicle use organizations, passengers, license plates, and drivers, wherein different types of nodes have edges, and passenger nodes have edges, wherein the weight of the edges is the number of times the edge relationship between nodes appears in historical vehicle use data, and the vehicle use relationship characteristics are feature vectors that characterize the association relationship between nodes in the heterogeneous network;
[0020] A prediction unit is used to input the vehicle usage relationship characteristics and the multiple vehicle usage demands into a preset deep learning model for prediction, so as to obtain available vehicle resources at a future moment, wherein the preset deep learning model is obtained by training the historical vehicle usage data, and the future moment is the next moment of the current moment.
[0021] In a possible implementation, the historical vehicle usage data includes a plurality of different vehicle dispatching tasks, and the vehicle resource prediction system further includes a construction unit, and the construction unit is specifically configured as follows:
[0022] Determine the factors that affect vehicle resources from multiple different vehicle dispatching tasks, and determine the correlation between factors of the same type and between factors of different types;
[0023] Determine the node type according to different types of factors, and determine the connection relationship between the nodes according to the association relationship between factors of the same type and the association relationship between factors of different types, wherein the node types include vehicle-using organizations, passengers, license plate numbers, and drivers;
[0024] The topological structure of the heterogeneous network is constructed according to the node types and the connection relationships between the nodes. The topological structure has four types of nodes, including vehicle-using organizations, passengers, license plates and drivers. There are edges between different types of nodes, and there are edges between passenger nodes. The weights of the edges are set to: the number of occurrences of the association relationships corresponding to the node edges in multiple different vehicle dispatching tasks.
[0025] In a possible implementation, the heterogeneous network includes a plurality of vehicle-using organization nodes, a plurality of passenger nodes, a plurality of license plate number nodes, and a plurality of driver nodes.
[0026] There are homogeneous edges between the passenger nodes, there are heterogeneous edges between the vehicle-using organization node and the passenger node, there are heterogeneous edges with a many-to-many relationship between the passenger node and the license plate number node, and there are heterogeneous edges with a many-to-many relationship between the license plate number node and the driver node.
[0027] In a possible implementation, the vehicle resource prediction system further includes a generation unit, and the generation unit is specifically configured as follows:
[0028] In the heterogeneous network, a plurality of node sequences are generated by randomly walking nodes;
[0029] Filtering at least one target node sequence from the multiple node sequences according to the edge weights in the node sequences;
[0030] After unifying the dimensions of the nodes in the target node sequence, a feature vector corresponding to the target node sequence is extracted, and the feature vector corresponding to the target node sequence is used as the vehicle use relationship feature.
[0031] A third aspect of the present application provides an electronic device, comprising at least one processor and a memory connected to the processor, wherein:
[0032] The memory is used to store computer programs;
[0033] The processor is used to execute the computer program so that the electronic device can implement the vehicle resource prediction method of the above-mentioned first aspect or any implementation method of the first aspect.
[0034] The fourth aspect of the present application provides a computer program product, including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements the vehicle resource prediction method of the above-mentioned first aspect or any implementation method of the first aspect.
[0035] By means of the above technical scheme, the present application provides a method, system, electronic device and computer program product for predicting vehicle resources. The method connects multiple factors involved in vehicle use, including vehicle use organizations, passengers, license plate numbers and drivers, through the topological structure of a heterogeneous network. The heterogeneous network includes four types of nodes: vehicle use organizations, passengers, license plate numbers and drivers, and there are edges between different types of nodes. There are edges between passenger nodes. The weight of the edge is the number of times the edge relationship between nodes appears in the historical vehicle use data. The corresponding vehicle use relationship features are generated according to the topological structure of the heterogeneous network. The vehicle use relationship features can represent the feature vector of the association relationship between nodes in the heterogeneous network. The vehicle use relationship features and multiple vehicle use demands input by the user end are input into a preset deep learning model trained based on historical vehicle use data for prediction, so that available vehicle resources at future times can be obtained. Since the consumption of vehicle resources is not only due to vehicle factors such as vehicle driving time and vehicle status, but may also involve other factors, the vehicle relationship features generated by the topological structure based on the heterogeneous network introduced in this solution can provide business association relationships between vehicle-related entities, so that when predicting vehicle resources, not only the impact of the vehicle itself on vehicle resources is considered, but also the impact of vehicle organizations, passengers and drivers on vehicle resource consumption is considered. This allows the preset deep learning model to more comprehensively consider various factors affecting vehicle resources, no longer limited to the vehicle itself, and can effectively improve the prediction accuracy of vehicle available resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.
[0037] Figure 1 A flow chart of a method for predicting vehicle resources provided in an embodiment of the present application;
[0038] Figure 2 A schematic diagram of the structure of a heterogeneous network provided in an embodiment of the present application;
[0039] Figure 3 A flowchart of another method for predicting vehicle resources provided in an embodiment of the present application;
[0040] Figure 4 A schematic diagram of the structure of a vehicle resource prediction system provided in an embodiment of the present application;
[0041] Figure 5 A hardware structure block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0042] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.
[0043] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.
[0044] The terms "first", "second" etc. in the specification of the application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable in appropriate circumstances, and this is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.
[0045] With the orderly expansion of the enterprise's vehicle business, the dispatch and management of official vehicles have gradually become more perfect, and a smart logistics system has been developed to achieve the management of official vehicles. The smart logistics system includes the storage maintenance of official vehicles and drivers, the application submission and approval of application reasons, etc. Therefore, the smart logistics system generates a huge amount of vehicle application and use data. In order to meet the increasingly abundant vehicle demand, improve the efficiency of vehicle call-up and save vehicle waiting time, the smart logistics system will predict the vehicle resources in the future to determine the vehicle dispatch method in advance.
[0046] At present, vehicle resources are often predicted in advance based on vehicle factors such as vehicle travel time and vehicle status, so that vehicles can be dispatched in advance to meet the increasingly diverse vehicle needs. However, in vehicle business, various factors (such as vehicles and users) can also have business coordination, and vehicles and users, and users and users can have associations. Therefore, the prediction of vehicle resources based only on vehicle factors is not accurate.
[0047] In order to solve the above problems, the embodiment of the present application provides a method for predicting vehicle resources. The method introduces the business association relationship between entities related to the vehicle, so that when predicting vehicle resources, not only the impact of the vehicle itself on the vehicle resources is considered, but also various factors affecting vehicle resources can be considered more comprehensively, effectively improving the prediction accuracy of vehicle available resources. The vehicle resource prediction method of the embodiment of the present application is described in detail below in conjunction with the accompanying drawings.
[0048] Reference Figure 1 , Figure 1 A flow chart of a method for predicting vehicle resources provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, a vehicle resource prediction method provided in an embodiment of the present application may include step S10 and step S11, and these steps are described in detail below.
[0049] S10. Receive multiple car-using demands input by the user end and obtain the current car-using relationship characteristics. The car-using relationship characteristics are generated based on the topological structure of the heterogeneous network. The heterogeneous network includes four types of nodes, namely, car-using organizations, passengers, license plates, and drivers. There are edges between different types of nodes. There are edges between passenger nodes. The weight of the edge is the number of times the edge relationship between nodes appears in the historical car-using data. The car-using relationship characteristics are feature vectors that characterize the association relationship between nodes in the heterogeneous network.
[0050] Among them, the user end can be a device used by users to query vehicle resources. The user can be a passenger or a manager of the smart logistics system.
[0051] The vehicle demand can be a screening condition for vehicle resources when inquiring. Optionally, the vehicle demand may include the departure place, destination, vehicle start time, vehicle end time, vehicle dispatch reason, number of passengers and vehicle duration, etc. The vehicle demand can be directly input by the user end, or provided to the user end by the smart logistics system and selected by the user end.
[0052] The vehicle use relationship feature is a feature that characterizes the association relationship between the various entities that use the vehicle. In this embodiment, the vehicle use relationship feature can be generated based on the topological structure of a heterogeneous network. A heterogeneous network refers to a network with different forms of nodes, edges, or connection methods. The topological structure of a heterogeneous network has multiple nodes, and there are edges between the nodes. The nodes can represent different types of entities such as people, objects, and events, and the edges between the nodes can represent the association relationship between the entities. Since the association relationship between the various entities is complex, this embodiment uses a complex heterogeneous network to better express the association relationship between the entities.
[0053] The construction process of the heterogeneous network in this embodiment can be shown as steps 1 to 3:
[0054] Step 1: Determine the factors that affect vehicle resources from multiple different vehicle dispatching tasks, and determine the correlation between factors of the same type and the correlation between factors of different types;
[0055] Step 2: Determine the node type according to different types of factors, and determine the connection relationship between nodes according to the correlation between factors of the same type and the correlation between factors of different types. The node types include vehicle-using organizations, passengers, license plate numbers, and drivers.
[0056] Step 3: Construct the topological structure of the heterogeneous network based on the node types and the connection relationships between nodes. The topological structure has four types of nodes, including vehicle-using organizations, passengers, license plates, and drivers. There are edges between different types of nodes, and there are edges between passenger nodes. The weight of the edges is set to: the number of occurrences of the association relationship corresponding to the node edges in multiple different vehicle dispatching tasks.
[0057] Among them, since the smart car module in the smart logistics system is a management application module for the dispatch of official vehicles for each business department, the module has functions such as vehicle management, driver management, dispatch reason management, motor vehicle application, motor vehicle approval, dispatch management and trip registration. Therefore, for each dispatch task, the smart car module records in detail the basic information such as the applicant, application agency, departure point, destination point, start time of the vehicle, end time of the vehicle, dispatch reason, number of passengers, license plate number, driver, etc. This embodiment can construct a heterogeneous network based on the dispatch task data in a large number of historical records. The constructed heterogeneous network is mainly used to describe the association relationship between various factors in different dispatch tasks in the historical records. These association relationships include the association relationship between factors of the same type, such as the relationship between passenger A and passenger B, and can also include the association relationship between factors of different types, such as passenger A took a vehicle with license plate C, and there is a relationship between passenger A and license plate C. Various factors may be involved in different vehicle dispatching tasks. This embodiment determines the node type in the heterogeneous network according to the various factors, and the association relationship between the key subjects can be used as the basis for the edge connection between the nodes, and the weight of the edge connection can be set to: the number of times the association relationship corresponding to the node edge connection appears in multiple different vehicle dispatching tasks. For example, if passenger A takes the vehicle with license plate C 25 times, the edge connection weight between the passenger A node and the license plate C node can be set to 25.
[0058] like Figure 2 The structural diagram of the heterogeneous network shown in FIG. The heterogeneous network may contain four types of nodes, namely, vehicle-using organizations, passengers, license plates and drivers, including multiple vehicle-using organization nodes, multiple passenger nodes, multiple license plate number nodes and multiple driver nodes. Since there is an affiliation relationship between the passenger and the vehicle-using organization, there may be heterogeneous edges (edges between two nodes of different types) between the vehicle-using organization node and the passenger node; since passengers may have business dealings with each other or ride in the same vehicle, there may be homogeneous edges (edges between two nodes of the same type) between the passenger nodes; since each passenger may have multiple official vehicle inquiries or applications in the smart logistics system, the vehicles dispatched each time may be different, so there may be many-to-many heterogeneous edges between the passenger node and the license plate number node; since the driver may drive different vehicles in each dispatch, there may be many-to-many heterogeneous edges between the license plate number node and the driver node. In this embodiment, the generated heterogeneous network can be represented as G=(V, E), where V can represent a node set and E can represent an edge set.
[0059] After constructing the heterogeneous network, this embodiment can generate the vehicle usage relationship features according to the topological structure of the heterogeneous network. The generation process of the vehicle usage relationship features can be shown in steps 4 to 6:
[0060] Step 4: Generate multiple node sequences by randomly walking nodes in the heterogeneous network;
[0061] Step 5: Select at least one target node sequence from multiple node sequences according to the edge weights in the node sequence;
[0062] Step 6: After unifying the dimensions of the nodes in the target node sequence, extract the feature vector corresponding to the target node sequence, and use the feature vector corresponding to the target node sequence as the vehicle usage relationship feature.
[0063] Among them, in the topological structure of heterogeneous networks, it is possible to randomly move from an initial node to an adjacent node. After repeated multiple times (m times), a node sequence can be formed. The node sequence can be expressed as: V i ={v1, v2...v m Specifically, this embodiment uses the node2vec method to implement random walk nodes in a heterogeneous network. The node2vec method combines the characteristics of breadth-first search (BFS) and depth-first search (DFS). By controlling the parameters, the probability of the transfer function between nodes can be changed, thereby optimizing the node sequence.
[0064] After obtaining multiple node sequences, this embodiment needs to screen the node sequences and determine the node sequences that meet the user end from the multiple node sequences. Specifically, this embodiment can screen at least one target node sequence from multiple node sequences based on the edge weights in the node sequence. Since the edge weights are the number of times the association relationship occurs, this embodiment screens the node sequence based on the sum of the edge weights in the node sequence or the average value of the weights. The target node sequence represents the vehicle use agencies, vehicles, and drivers that users often use in multiple vehicle dispatch tasks recorded in the history, which means that the degree of association between the target node sequence and the user end is higher than that of other node sequences. After obtaining the target node sequence, this embodiment can apply an embedded learning algorithm to process the target node sequence, extract the feature vector corresponding to the target node sequence, and use it as a vehicle use relationship feature. Further, the feature vector can be expressed as: W={w1, w2…w n}.
[0065] This embodiment introduces the vehicle usage relationship feature, which fully considers the relationship between the various entities that affect vehicle resources, such as the business relationship between passengers. It can more accurately predict available vehicle resources and guide applicants to make reasonable vehicle resource applications based on changes in vehicles and drivers, making the vehicle usage process more efficient, avoiding idle vehicles and drivers and business waiting when there are no vehicles available, thereby improving utilization efficiency.
[0066] S11. Input the vehicle usage relationship characteristics and multiple vehicle usage demands into a preset deep learning model for prediction to obtain available vehicle resources at future moments. The preset deep learning model is trained by historical vehicle usage data, and the future moment is the next moment of the current moment.
[0067] Among them, since the changes in car-use tasks between different time periods have a certain correlation, the deep learning model used in this embodiment is the DeepAR prediction model. The DeepAR model is a deep learning model specifically used to process time series probability predictions, which can automatically learn complex patterns in the data and improve the accuracy of predictions. DeepAR is an optimization of recursive neural networks. It can combine multiple similar time series, learn the correlation characteristics in different time series through deep recursive neural networks, and use multiple or multiple target numbers to improve the overall prediction accuracy. DeepAR can supplement default values within the model to reduce the demand for data, as well as the early data regularization and cleaning work. Among them, the recursive neural network is a variant of the recursive artificial neural network. In the recursive artificial neural network, the connection between neurons forms a directed loop, so that the output depends not only on the current input, but also on the state of the neurons in the previous step.
[0068] The overall process of training the DeepAR prediction model in this embodiment can be described as follows:
[0069] The corresponding feature vector is constructed according to the vehicle demand in the historical vehicle usage data, which can be expressed as: U = {u1, u2…u p Combining the car use relationship features and multiple car use requirements as the input features of the DeepAR prediction model at the current moment, the input features can be expressed as: X t ={u1,u2……u p , w1, w2...w n}, and the corresponding license plate number and driver are used as labels, which can be expressed as: Y={y1,y2}.
[0070] Determine a division time z, use the z time to divide the historical vehicle data, obtain the historical vehicle data before z time and the historical vehicle data after z time, and use a sliding window with a window size of T to segment the historical vehicle data. The historical vehicle data before z time is used as the training data set, which can be expressed as: X train ={X t+1， X t+2 ...X t+T}, label: Y train ={Y t+T}; Taking the historical vehicle usage data after time z as the test data set, it can be expressed as: X test ={X t+z+1， X t+z+2 ...X t+z+T}, label: Y test ={Y t+z+T}. This embodiment constructs a time series based on the training data set, uses the feature vector before time T to predict the situation at time t+T, and continuously updates the neural network parameters in the DeepAR prediction model with the goal of maximizing the objective function (log-likelihood function) of the DeepAR prediction model. After the DeepAR prediction model is trained, the prediction accuracy of the DeepAR prediction model is tested using a test data set. When the prediction accuracy meets the requirements, the preset deep learning model of this embodiment is finally obtained. Furthermore, the smart logistics system can continuously collect vehicle usage data and continuously update the DeepAR prediction model to improve the accuracy of the DeepAR prediction model.
[0071] The available vehicle resources output by the preset deep learning model may be available resources that meet the user's vehicle demand, and the specific form of the available vehicle resources may be the proportion of idle vehicles that meet the user's vehicle demand. The prediction results based on the preset deep learning model output in this embodiment can show the user the degree of vehicle demand during the time period that meets the user's demand, so as to guide the user to submit the user application scientifically and reasonably. Furthermore, this embodiment may also use other deep learning models for prediction.
[0072] The embodiment of the present application provides a method for predicting vehicle resources. The method connects multiple factors involved in vehicle use, including vehicle use organizations, passengers, license plate numbers, and drivers, through the topological structure of a heterogeneous network. The heterogeneous network includes four types of nodes: vehicle use organizations, passengers, license plate numbers, and drivers. There are edges between different types of nodes, and there are edges between passenger nodes. The weight of the edge is the number of times the edge relationship between nodes appears in the historical vehicle use data. The corresponding vehicle use relationship features are generated according to the topological structure of the heterogeneous network. The vehicle use relationship features can represent the feature vector of the association relationship between nodes in the heterogeneous network. The vehicle use relationship features and multiple vehicle use demands input by the user end are input into a preset deep learning model trained based on historical vehicle use data for prediction, so that available vehicle resources at future times can be obtained. Therefore, this embodiment uses a heterogeneous network to model the smart car call management, draws an application relationship diagram, and vectorizes the abstract application relationship network through the node2vec method to form car relationship characteristics. Based on the car relationship characteristics and the user needs in the car application approval form filled out by the user, the DeepAR prediction model is used to predict the car resource situation in the next time window.
[0073] Since the consumption of vehicle resources is not only due to vehicle factors such as vehicle driving time and vehicle status, but may also involve other factors, the vehicle relationship features generated by the topological structure based on the heterogeneous network introduced in this solution can provide business association relationships between vehicle-related entities, so that when predicting vehicle resources, not only the impact of the vehicle itself on vehicle resources is considered, but also the impact of vehicle organizations, passengers and drivers on vehicle resource consumption is considered. This allows the preset deep learning model to more comprehensively consider various factors affecting vehicle resources, no longer limited to the vehicle itself, and can effectively improve the prediction accuracy of vehicle available resources.
[0074] Here, an embodiment is provided to briefly illustrate the above overall technical solution. Figure 3 The process diagram is shown.
[0075] S30, determining various factors affecting vehicle resources and the correlation between the factors from multiple vehicle dispatching tasks in the historical vehicle use data, and generating a heterogeneous network topology structure representing the vehicle use correlation based on the correlation between the factors, wherein the factors include the vehicle use organization, the passenger, the license plate number, and the driver;
[0076] S31, using the node2vec method to generate vehicle usage relationship features that represent node relationship vectors from heterogeneous network topology structures;
[0077] S32, selecting at least one vehicle use demand from the vehicle use demands input by the user at the current moment, and generating basic application features, wherein the vehicle use demand includes a departure place, a destination, a vehicle use start time, a vehicle use end time, a vehicle use reason, a number of passengers, and a vehicle use duration, etc.;
[0078] S33, constructing a DeepAR prediction model, and training the DeepAR prediction model to obtain a preset deep learning model;
[0079] S34. Input the vehicle usage relationship characteristics and basic application characteristics at the current moment into a preset deep learning model, and use the preset deep learning model to predict the available vehicle resources at a future moment, where the future moment is the next moment of the current moment.
[0080] A method for predicting vehicle resources provided in an embodiment of the present application is introduced above, and a system applying the above method for predicting vehicle resources will be introduced below.
[0081] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of a vehicle resource prediction system provided in an embodiment of the present application. Figure 4 As shown, the vehicle resource prediction system includes:
[0082] The receiving unit 100 is used to receive multiple vehicle use demands input by the user end and obtain the current vehicle use relationship characteristics. The vehicle use relationship characteristics are generated based on the topological structure of the heterogeneous network. The heterogeneous network includes four types of nodes, namely, vehicle use organizations, passengers, license plates, and drivers. There are edges between different types of nodes. There are edges between passenger nodes. The weight of the edge is the number of times the edge relationship between nodes appears in the historical vehicle use data. The vehicle use relationship characteristics are feature vectors that characterize the association relationship between nodes in the heterogeneous network.
[0083] The prediction unit 110 is used to input the vehicle usage relationship characteristics and multiple vehicle usage demands into a preset deep learning model for prediction, and obtain the available vehicle resources at a future moment. The preset deep learning model is trained by historical vehicle usage data, and the future moment is the next moment of the current moment.
[0084] In a possible implementation, the historical vehicle usage data includes a plurality of different vehicle dispatching tasks, and the vehicle resource prediction system further includes a construction unit, and the construction unit is specifically configured as follows:
[0085] Determine the factors that affect vehicle resources from multiple different vehicle dispatching tasks, and determine the correlation between factors of the same type and between factors of different types;
[0086] Determine the node type according to different types of factors, and determine the connection relationship between nodes according to the correlation between factors of the same type and the correlation between factors of different types. The node types include vehicle-using organizations, passengers, license plate numbers, and drivers.
[0087] The topological structure of the heterogeneous network is constructed according to the node types and the connection relationships between the nodes. The topological structure has four types of nodes, including vehicle-using organizations, passengers, license plates and drivers. There are edges between different types of nodes, and there are edges between passenger nodes. The weight of the edges is set to: the number of occurrences of the association relationship corresponding to the node edges in multiple different vehicle dispatching tasks.
[0088] In a possible implementation, the heterogeneous network includes multiple vehicle-using organization nodes, multiple passenger nodes, multiple license plate number nodes, and multiple driver nodes.
[0089] There are homogeneous edges between the passenger nodes, heterogeneous edges between the vehicle-using organization node and the passenger node, many-to-many heterogeneous edges between the passenger node and the license plate number node, and many-to-many heterogeneous edges between the license plate number node and the driver node.
[0090] In a possible implementation, the vehicle resource prediction system further includes a generation unit, and the generation unit is specifically configured as follows:
[0091] In heterogeneous networks, multiple node sequences are generated by randomly walking nodes;
[0092] Filtering at least one target node sequence from multiple node sequences according to the edge weights in the node sequences;
[0093] After unifying the dimensions of the nodes in the target node sequence, the feature vector corresponding to the target node sequence is extracted, and the feature vector corresponding to the target node sequence is used as the vehicle usage relationship feature.
[0094] The present application also provides an electronic device in an embodiment. Figure 5 As shown, it shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiment of the present application. The electronic device in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 5 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.
[0095] like Figure 5As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 to a random access memory (RAM) 503. When the electronic device is powered on, various programs and data required for the operation of the electronic device are also stored in the RAM 503. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.
[0096] Typically, the following devices may be connected to the I / O interface 505: an input device 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 508 including, for example, a memory card, a hard disk, etc.; and a communication device 509. The communication device 509 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although Figure 5 An electronic device having various devices is shown, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.
[0097] An embodiment of the present application also provides a computer program product including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any of the vehicle resource prediction methods provided in the embodiments of the present application.
[0098] A computer-readable storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any vehicle resource prediction method provided in an embodiment of the present application.
[0099] It should also be noted that the system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the system embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.
[0100] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a magnetic disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.
[0101] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.
[0102] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a training device, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.
[0103] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0104] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the types, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.
[0105] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.
Claims
1. A method for predicting vehicle resources, characterized in that: The vehicle resource prediction method comprises: Receive multiple vehicle use demands input by a user end and obtain current vehicle use relationship features, wherein the vehicle use relationship features are generated based on the topological structure of a heterogeneous network, wherein the heterogeneous network includes four types of nodes, namely, vehicle use organizations, passengers, license plates, and drivers, wherein different types of nodes have edges, and passenger nodes have edges, wherein the weight of the edges is the number of times the edge relationship between nodes appears in historical vehicle use data, and the vehicle use relationship features are feature vectors that characterize the node association relationship in the heterogeneous network; The vehicle usage relationship characteristics and the multiple vehicle usage demands are input into a preset deep learning model for prediction to obtain available vehicle resources at a future moment, wherein the preset deep learning model is trained by the historical vehicle usage data, and the future moment is the next moment of the current moment.
2. The vehicle resource prediction method according to claim 1, characterized in that: The historical vehicle usage data includes a plurality of different vehicle dispatching tasks, and the construction process of the heterogeneous network includes: Determine the factors that affect vehicle resources from multiple different vehicle dispatching tasks, and determine the correlation between factors of the same type and between factors of different types; Determine the node type according to different types of factors, and determine the connection relationship between the nodes according to the association relationship between factors of the same type and the association relationship between factors of different types, wherein the node types include vehicle-using organizations, passengers, license plate numbers, and drivers; The topological structure of the heterogeneous network is constructed according to the node types and the connection relationships between the nodes. The topological structure has four types of nodes, including vehicle-using organizations, passengers, license plates and drivers. There are edges between different types of nodes, and there are edges between passenger nodes. The weights of the edges are set to: the number of occurrences of the association relationships corresponding to the node edges in multiple different vehicle dispatching tasks.
3. The vehicle resource prediction method according to claim 1, characterized in that: The heterogeneous network includes a plurality of vehicle-using organization nodes, a plurality of passenger nodes, a plurality of license plate number nodes and a plurality of driver nodes. There are homogeneous edges between the passenger nodes, there are heterogeneous edges between the vehicle-using organization node and the passenger node, there are heterogeneous edges with a many-to-many relationship between the passenger node and the license plate number node, and there are heterogeneous edges with a many-to-many relationship between the license plate number node and the driver node.
4. The vehicle resource prediction method according to claim 1, characterized in that: The generation process of the vehicle usage relationship feature is as follows: In the heterogeneous network, a plurality of node sequences are generated by randomly walking nodes; Filtering at least one target node sequence from the multiple node sequences according to the edge weights in the node sequences; After unifying the dimensions of the nodes in the target node sequence, a feature vector corresponding to the target node sequence is extracted, and the feature vector corresponding to the target node sequence is used as the vehicle use relationship feature.
5. A vehicle resource prediction system, characterized in that: The vehicle resource prediction system comprises: A receiving unit, used to receive multiple vehicle use demands input by a user terminal and obtain current vehicle use relationship characteristics, wherein the vehicle use relationship characteristics are generated based on the topological structure of a heterogeneous network, wherein the heterogeneous network includes four types of nodes, namely, vehicle use organizations, passengers, license plates, and drivers, wherein different types of nodes have edges, and passenger nodes have edges, wherein the weight of the edges is the number of times the edge relationship between nodes appears in historical vehicle use data, and the vehicle use relationship characteristics are feature vectors that characterize the association relationship between nodes in the heterogeneous network; A prediction unit is used to input the vehicle usage relationship characteristics and the multiple vehicle usage demands into a preset deep learning model for prediction, so as to obtain available vehicle resources at a future moment, wherein the preset deep learning model is obtained by training the historical vehicle usage data, and the future moment is the next moment of the current moment.
6. The vehicle resource prediction system according to claim 5, characterized in that: The historical vehicle usage data includes a plurality of different vehicle dispatching tasks, and the vehicle resource prediction system further includes a construction unit, and the construction unit is specifically configured as follows: Determine the factors that affect vehicle resources from multiple different vehicle dispatching tasks, and determine the correlation between factors of the same type and between factors of different types; Determine the node type according to different types of factors, and determine the connection relationship between the nodes according to the association relationship between factors of the same type and the association relationship between factors of different types, wherein the node types include vehicle-using organizations, passengers, license plate numbers, and drivers; The topological structure of the heterogeneous network is constructed according to the node types and the connection relationships between the nodes. The topological structure has four types of nodes, including vehicle-using organizations, passengers, license plates and drivers. There are edges between different types of nodes, and there are edges between passenger nodes. The weights of the edges are set to: the number of occurrences of the association relationships corresponding to the node edges in multiple different vehicle dispatching tasks.
7. The vehicle resource prediction system according to claim 5, characterized in that: The heterogeneous network includes a plurality of vehicle-using organization nodes, a plurality of passenger nodes, a plurality of license plate number nodes and a plurality of driver nodes. There are homogeneous edges between the passenger nodes, there are heterogeneous edges between the vehicle-using organization node and the passenger node, there are heterogeneous edges with a many-to-many relationship between the passenger node and the license plate number node, and there are heterogeneous edges with a many-to-many relationship between the license plate number node and the driver node.
8. The vehicle resource prediction system according to claim 5, characterized in that: The vehicle resource prediction system further includes a generation unit, which is specifically configured as follows: In the heterogeneous network, a plurality of node sequences are generated by randomly walking nodes; Filtering at least one target node sequence from the multiple node sequences according to the edge weights in the node sequences; After unifying the dimensions of the nodes in the target node sequence, a feature vector corresponding to the target node sequence is extracted, and the feature vector corresponding to the target node sequence is used as the vehicle use relationship feature.
9. An electronic device, characterized in that: The method comprises at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the electronic device can implement the vehicle resource prediction method as described in any one of claims 1 to 4.
10. A computer program product, characterized in that It includes computer-readable instructions, and when the computer-readable instructions are executed on an electronic device, the electronic device implements the vehicle resource prediction method as described in any one of claims 1 to 4.