A vehicle intelligent scheduling method and system based on vehicle networking

Abstract

The application provides a vehicle intelligent scheduling method and system based on Internet of Vehicles, and relates to the technical field of intelligent transportation. The method comprises the following steps: acquiring real-time data and historical data; dynamically evaluating the reliability of each unfinished order in returning the vehicle according to the plan to obtain a weighted planned return quantity; obtaining a temporary pick-up demand prediction value through a time series prediction model and converting the temporary pick-up demand prediction value into a temporary return quantity prediction value; fusing the reserved pick-up demand and the predicted pick-up demand to obtain a total pick-up demand, fusing the weighted planned return quantity and the temporary return quantity prediction value to obtain a total return supply, and calculating a net demand; and constructing a multi-objective scheduling optimization model to solve a scheduling instruction. The application realizes high-precision prediction and low-cost scheduling of vehicle supply and demand in the Internet of Vehicles environment by deeply fusing individual plan commitment and group statistical law.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, and in particular to a vehicle intelligent scheduling method and system based on the Internet of Vehicles. Background Technology

[0002] In the context of the Internet of Vehicles (IoV), car-sharing and car rental services exhibit a typical "tidal" imbalance between vehicle supply and demand in terms of time and space: during peak hours and holidays in popular areas, some service points are experiencing a shortage of vehicles, while other service points have a large number of idle vehicles. Existing dispatch systems typically employ statistical forecasting methods based on historical data, such as time series analysis or simple machine learning models, to predict future pick-up and drop-off demand at various service points and then formulate dispatch strategies.

[0003] However, such methods can only handle macro-level probabilistic patterns. When users' actual vehicle pick-up and drop-off behavior deviates from historical statistical distributions, there will be a large discrepancy between the prediction results and the actual supply and demand, resulting in delayed dispatch instructions and increased vehicle empty mileage. They cannot achieve forward-looking and accurate dispatch in the dynamic and uncertain Internet of Vehicles environment. Summary of the Invention

[0004] In view of the technical problems existing in the prior art, the present invention provides a vehicle intelligent scheduling method and system based on the Internet of Vehicles.

[0005] The first aspect of the present invention provides a vehicle intelligent scheduling method based on the Internet of Vehicles, comprising: Obtain the number of available vehicles at each location at the current time, the vehicle usage plans and real-time GPS trajectories of unfinished orders, the historical temporary vehicle pick-up demand sequence and external characteristics, and the known external characteristics for future periods; For each unfinished order, the reliability of returning the car as planned is dynamically assessed based on its usage plan and real-time GPS trajectory. The weighted planned return volume for each time period and each network point is calculated using this reliability as the weight. The historical temporary vehicle retrieval demand sequence and external features are input into a pre-trained time series prediction model to obtain the predicted values ​​of temporary vehicle retrieval demand for each location in the future time period. The predicted value of temporary vehicle pick-up demand is input into the temporary vehicle return conversion model based on the joint distribution of rental period and return location to obtain the predicted value of temporary vehicle return volume for each location in the future time period. The total vehicle pick-up demand is obtained by adding the reserved vehicle pick-up demand to the predicted temporary vehicle pick-up demand, and the total vehicle return supply is obtained by adding the weighted planned vehicle return quantity to the predicted temporary vehicle return quantity. Then, the net demand of each outlet in the future for each time period is calculated. Based on the net demand, available vehicles, network capacity, and scheduling cost, a multi-objective scheduling optimization model is constructed and solved to obtain the number of vehicles dispatched from each network point to each network point and the number of dispatches, which are then output as scheduling instructions.

[0006] Furthermore, the dynamic assessment of the reliability of returning each incomplete order as scheduled specifically includes: Obtain basic order characteristics, user profile characteristics, and static environmental characteristics at the time of order generation; The above features are input into a pre-trained classification model to obtain the initial credibility score of the order; During vehicle use, the real-time GPS trajectory sequence of the vehicle is acquired, and the direction consistency factor, abnormal stop factor and path deviation factor are extracted. The factors are then time-weighted and modulated according to the vehicle use progress to obtain dynamic adjustment coefficients. The initial confidence score is multiplied by the dynamic adjustment coefficient, and then subjected to amplitude limiting to obtain the real-time confidence score at the current moment. The real-time reliability at the current moment is calculated using the following formula: ; in, For orders The initial credibility score, For the first A dynamic adjustment coefficient.

[0007] Furthermore, the classification model includes any one or more of the following: an ensemble learning model based on decision trees, a logistic regression model, and a support vector machine model; the time series prediction model includes any one or more of the following: a long short-term memory network, a gated recurrent unit, a recurrent neural network, and a support vector regression model.

[0008] Furthermore, the calculation method for the predicted value of temporary vehicle returns is as follows: Get the number of orders from each pick-up location, each rental period, and each return location from the historical order data; A joint probability table of rental period and return location was obtained, where each element represents the distance from the pick-up point to the drop-off point. Departure, rental period is During a specific time period, or at a branch probability ; For any future time period Temporary car retrieval demand forecast Multiply this by the corresponding joint probability distribution, and sum over all pick-up times and pick-up locations to obtain the probability for any future time period. At any point The predicted value of temporary car return volume; The predicted value of temporary vehicle return volume is calculated using the following formula: ; in, For from the outlet Dispatch a vehicle to the branch. The cost, This represents the total number of time periods to be predicted.

[0009] Furthermore, the construction and solution of the multi-objective scheduling optimization model specifically includes: The objective function is to minimize the sum of the weighted scheduling mileage cost and the weighted scheduling count cost, where the mileage cost is weighted. Weighted by frequency cost satisfy ; Set constraints for vehicle flow conservation, available vehicles, network capacity, the relationship between scheduling frequency and vehicle number, state transition equations, and initial conditions. The optimization model is solved using a mixed-integer linear programming solver or a heuristic algorithm to obtain the network points for each time period. Dispatch to branch Number of vehicles and number of scheduling ; Only execute the scheduling instructions for the current time period, and solve the problem on a rolling basis at the next decision time; The specific form of the objective function is as follows: ; in, For from the outlet Dispatch a vehicle to the branch. The cost, This represents the total number of time periods to be predicted.

[0010] Furthermore, it also includes a closed-loop self-learning step: Collect actual operational data, including actual vehicle pick-up demand, actual vehicle return volume, actual return time and location for each order, and execution results of dispatch instructions at each time period and location. Calculate the prediction error of temporary vehicle retrieval demand and feed the error back into the time series prediction model for retraining; Calculate the Brier score between the credibility of each order and whether the vehicle was actually returned as planned, and feed this score back to the classification model for incremental updates, while adjusting the weights of the direction consistency factor, abnormal stay factor, and path deviation factor. The mileage cost weight in the multi-objective scheduling optimization model is adaptively adjusted based on the actual demand satisfaction rate and the occurrence rate of peak traffic. Weighted by frequency cost .

[0011] A second aspect of the present invention provides a vehicle intelligent dispatching system based on the Internet of Vehicles, comprising: The data acquisition module is used to acquire the number of available vehicles at each network point at the current time, the vehicle usage plans and real-time GPS trajectories of unfinished orders, the historical temporary vehicle pick-up demand sequence and external characteristics, and the known external characteristics of future time periods; The credibility dynamic assessment module is used to dynamically assess the credibility of returning the car as planned for each incomplete order based on its car usage plan and real-time GPS trajectory, and calculate the weighted planned car return volume for each time period and each network point in the future using the credibility as the weight. The prediction module is used to input the historical temporary vehicle retrieval demand sequence and external features into a pre-trained time series prediction model to obtain the predicted values ​​of temporary vehicle retrieval demand for each location in the future time period. The temporary vehicle return statistics module is used to input the predicted value of temporary vehicle pick-up demand into the temporary vehicle return conversion model based on the joint distribution of rental period and return location to obtain the predicted value of temporary vehicle return volume for each location in the future. The supply and demand fusion calculation module is used to add the reserved car pick-up demand to the predicted value of the temporary car pick-up demand to obtain the total car pick-up demand, and add the weighted planned car return quantity to the predicted value of the temporary car return quantity to obtain the total car return supply, and then calculate the net demand of each network point in the future for each time period. The vehicle scheduling module is used to construct and solve a multi-objective scheduling optimization model based on the net demand, the number of available vehicles, the network capacity, and the scheduling cost, and to obtain the number of vehicles dispatched from each network to each network and the number of scheduling operations, which are then output as scheduling instructions.

[0012] A third aspect of the present invention provides an electronic device including a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps of the vehicle intelligent scheduling method based on the Internet of Vehicles as described in the first aspect of the present invention.

[0013] A fourth aspect of the present invention provides a computer-readable storage medium having a program stored thereon that, when executed by a processor, implements the steps of the vehicle intelligent scheduling method based on the Internet of Vehicles as described in the first aspect of the present invention.

[0014] A fifth aspect of the present invention provides a computer program product comprising software code, wherein the program in the software code performs the steps of the vehicle intelligent scheduling method based on the Internet of Vehicles as described in the first aspect of the present invention.

[0015] Compared with existing technologies, the vehicle intelligent scheduling method and system based on the Internet of Vehicles (IoV) provided by this invention has the following beneficial effects: By acquiring and dynamically evaluating the reliability of scheduled vehicle returns for each unfinished order, and simultaneously obtaining the predicted value of temporary vehicle retrieval demand through a time series prediction model and converting it into the predicted value of temporary vehicle return volume, the total retrieval demand is obtained by adding the scheduled retrieval demand and the temporary retrieval demand, and the total return supply is obtained by adding the weighted planned return volume and the predicted value of temporary return volume. This allows for the calculation of net demand and the construction of a multi-objective scheduling optimization model. By organically integrating individual "planned commitments" with group "statistical regularities," the accuracy of supply and demand perception is improved from the source, enabling scheduling instructions to simultaneously respond to deterministic orders and probabilistic temporary demands. This effectively alleviates the spatiotemporal imbalance problem of vehicle supply and demand, reduces scheduling mileage and the number of scheduling operations, and improves vehicle utilization and user experience. Attached Figure Description

[0016] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.

[0017] Figure 1 A flowchart of a vehicle intelligent scheduling method based on the Internet of Vehicles provided in Embodiment 1 of the present invention; Figure 2 This is a block diagram of a vehicle intelligent dispatching system based on the Internet of Vehicles provided in Embodiment 2 of the present invention. Detailed Implementation

[0018] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form as well. Furthermore, it should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0020] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0021] All data acquisition in this embodiment is carried out in accordance with laws and regulations and with user consent, and the data is used legally.

[0022] Example 1 like Figure 1 This embodiment provides a vehicle intelligent scheduling method based on the Internet of Vehicles, including: S1. Obtain the number of available vehicles at each branch at the current time, the vehicle usage plans and real-time GPS trajectories of unfinished orders, the historical temporary vehicle pick-up demand sequence and external characteristics, and the known external characteristics for future periods.

[0023] Specifically, the current moment is defined as the decision point. The prediction time window length is Future Time Index ,in To predict the total number of time periods, express Time. Assume the total number of outlets is... outlets The maximum capacity is From the outlet Dispatch a vehicle to the branch. The cost is . express Time Network Points The number of available vehicles. The data obtained in this step includes: the current number of available vehicles at each service point. Each unfinished order ( , The car rental plan (planned return time) is based on the number of currently active orders. and planned car return locations ) and real-time GPS trajectory sequences (Including longitude, latitude, speed, and direction); Temporary vehicle retrieval demand sequence for K historical time periods. and corresponding external features Known external characteristics for the next T time periods External characteristics include time characteristics (time of day, day of the week, holidays), weather characteristics, and branch attributes.

[0024] S2. For each unfinished order, based on its usage plan and real-time GPS trajectory, dynamically assess its reliability of returning the vehicle as planned, and use this reliability as a weight to calculate the weighted planned return volume for each time period and each network point in the future.

[0025] This step employs a two-layer architecture of "static baseline + dynamic tracking". The first layer (static initial assessment): When an order is generated, basic order features (rental duration, planned return location type), user profile features (historical punctuality rate, historical average time deviation, membership level, credit score, etc.), and environmental static features (whether the pick-up / drop-off time is day / night, whether it's a holiday, etc.) are obtained. These features are then input into a pre-trained binary classification model (logistic regression or XGBoost, etc.) to output an initial confidence score. .

[0026] The second layer (dynamic trip tracking): After the vehicle trip begins, it tracks the vehicle's real-time GPS trajectory sequence. Reliability is updated dynamically. Three basic adjustment coefficients are defined: Directional consistency factor Calculate the angle between the vehicle's direction and the planned network point direction. ,like (e.g., 30°) and vehicle speed If the value is 1, then take 1.1; otherwise, take 1.0.

[0027] Abnormal retention factor If the continuous stay time If the vehicle is not within the electronic fence, the value is 0.7; otherwise, the value is 1.0.

[0028] Path deviation factor Let the straight-line distance from the vehicle to the planned network point be... Distance threshold is ,like If the value is 0.5, then take 0.5; otherwise, take 1.0. This is the start time for vehicle pickup of the order. The planned return time for the order. The allowable deviation range gradually decreases over time.

[0029] Time-weighted modulation is applied to the above basic adjustment coefficients: ,in The decay index (usually taken as 1.5~2.0) Initial stage of car use , (No impact); close to the time of returning the car , (Complete impact).

[0030] The formula for updating the real-time credibility at the current moment is: ; in As an indicator function, when the scheduled return time falls within the time period The value is 1 if the time is within the range, and 0 otherwise.

[0031] S3. Input the historical temporary vehicle retrieval demand sequence and external features into a pre-trained time series prediction model to obtain the predicted values ​​of temporary vehicle retrieval demand for each location in the future.

[0032] Specifically, a Long Short-Term Memory (LSTM) network is used as the time series prediction model (gated recurrent units, recurrent neural networks, or support vector regression models can also be used). The input is the sequence of temporary vehicle retrieval demand for each network point over K historical time periods. External characteristics of the corresponding time period and known external characteristics for the next T time periods. The model outputs predicted temporary car retrieval demand for the next T time periods: ; The prediction model expression is: .

[0033] S4. Input the predicted value of temporary vehicle pick-up demand into the temporary vehicle return conversion model based on the joint distribution of rental period and return location to obtain the predicted value of temporary vehicle return volume for each location in the future.

[0034] For the predicted temporary car retrieval demand Since there is no specific order information, it is necessary to predict the return distribution based on historical statistical patterns. First, a joint distribution probability table of rental period and return location is established: This is done by statistically analyzing historical orders from pick-up locations... Departure, rental period is At what time period, and at the branch probability The calculation formula is: ; satisfy , Determined based on actual orders.

[0035] The number of car returns resulting from temporary car pick-up requests is: ; This formula distributes all temporary car pick-up requests for all pick-up times and locations to each return location and time period according to a joint distribution probability.

[0036] S5. Add the reserved car pick-up demand to the predicted value of the temporary car pick-up demand to obtain the total car pick-up demand, add the weighted planned car return quantity to the predicted value of the temporary car return quantity to obtain the total car return supply, and then calculate the net demand of each outlet for each time period in the future.

[0037] Let the demand for scheduled car pickup be (Given deterministic orders), the total vehicle pickup demand is: ; Total rental car supply is: ; Net demand is defined as the difference between total vehicle pickup demand and (current available vehicles and total vehicle return supply), which will be reflected in the constraints of the subsequent optimization model.

[0038] S6. Based on the net demand, number of available vehicles, network capacity and scheduling cost, construct and solve a multi-objective scheduling optimization model to obtain the number of vehicles dispatched from each network to each network and the number of scheduling operations, which are then output as scheduling instructions.

[0039] This step constructs a mixed-integer linear programming (MILP) model. First, define the variables: Time period Internal outlets Dispatch to network point The number of vehicles; Time period Internal outlets Dispatch to network point Number of scheduling ( );

[0040] : Slack variables Indicates time period Unmet vehicle shortage (positive indicates shortage, negative indicates surplus). Indicates time period The number of over-capacity vehicles that have not been removed from the inventory list. ; Time period At the start (before dispatch) of the network point Number of vehicles, initial state: .

[0041] The objective function is to minimize the sum of the weighted scheduling mileage cost and the scheduling count cost: ; in , , where represents the weights of mileage cost and scheduling frequency cost, respectively.

[0042] The constraints include: (1) Conservation of vehicle flow and satisfaction of net demand ; The first term on the left represents the net inflow resulting from scheduling, and the second term represents the original net demand (positive indicates a shortage of vehicles, negative indicates a surplus). Slack variables allow for unmet demand or surplus vehicles that have not been scheduled out.

[0043] (2) Vehicle constraints available ; (3) Capacity constraints The number of vehicles after dispatch (before natural vehicle pick-up and return) must be within the capacity range: ; (4) The number of dispatches is related to the number of vehicles. ; ; ; in, It is a sufficiently large constant (such as the total number of vehicles).

[0044] (5) Scheduling resource constraints (this constraint may be disregarded if appropriate) The number of dispatch tasks executed simultaneously within the same time period shall not exceed the number of available dispatchers / dispatch vehicles: ; (6) State transition equation ; (7) Initial conditions ; (8) Variable nonnegation and integer constraints

[0045] The solution can be obtained using heuristic algorithms such as genetic algorithms and simulated annealing, or by using commercial solvers (such as Gurobi and CPLEX) to obtain an exact solution. In practice, only one method is used. The scheduling instructions (i.e., the current time period) are given; the solution is recalculated at the next decision moment to achieve rolling optimization and ensure real-time performance.

[0046] Preferably, this embodiment further includes a closed-loop self-learning step: Collect actual operational data: actual vehicle pick-up demand at each location during different time periods. Actual number of returned vehicles The actual return time and location of each order, and the results of the dispatch instructions.

[0047] Calculate the forecast error of temporary vehicle retrieval demand The results are then fed back to the time series prediction model for retraining.

[0048] Calculate the Brier score (in (This indicates that the car was actually returned as planned), which is fed back to the classification model for incremental updates, while the weights of the direction consistency factor, abnormal stay factor, and path deviation factor are adjusted.

[0049] Based on demand fulfillment rate The occurrence rate of traffic spikes (the percentage of time periods when the number of vehicles exceeds the capacity of the service area) is used to adaptively adjust the weight coefficients in the objective function. and .

[0050] Through the above steps, the vehicle intelligent scheduling method based on the Internet of Vehicles provided in this embodiment deeply integrates individual "plan commitments" with group "statistical patterns", realizing high-precision prediction and low-cost scheduling of vehicle supply and demand in the Internet of Vehicles environment.

[0051] Example 2 like Figure 2 This embodiment provides a vehicle intelligent dispatching system based on the Internet of Vehicles, including: The data acquisition module is used to acquire the number of available vehicles at each network point at the current time, the vehicle usage plans and real-time GPS trajectories of unfinished orders, the historical temporary vehicle pick-up demand sequence and external characteristics, and the known external characteristics of future time periods; The credibility dynamic assessment module is used to dynamically assess the credibility of returning the car as planned for each incomplete order based on its car usage plan and real-time GPS trajectory, and calculate the weighted planned car return volume for each time period and each network point in the future using the credibility as the weight. The prediction module is used to input the historical temporary vehicle retrieval demand sequence and external features into a pre-trained time series prediction model to obtain the predicted values ​​of temporary vehicle retrieval demand for each location in the future time period. The temporary vehicle return statistics module is used to input the predicted value of temporary vehicle pick-up demand into the temporary vehicle return conversion model based on the joint distribution of rental period and return location to obtain the predicted value of temporary vehicle return volume for each location in the future. The supply and demand fusion calculation module is used to add the reserved car pick-up demand to the predicted value of the temporary car pick-up demand to obtain the total car pick-up demand, and add the weighted planned car return quantity to the predicted value of the temporary car return quantity to obtain the total car return supply, and then calculate the net demand of each network point in the future for each time period. The vehicle scheduling module is used to construct and solve a multi-objective scheduling optimization model based on the net demand, the number of available vehicles, the network capacity, and the scheduling cost, and to obtain the number of vehicles dispatched from each network to each network and the number of scheduling operations, which are then output as scheduling instructions.

[0052] Preferably, the system may further include a closed-loop self-learning module, used to collect actual operational data, calculate prediction error and reliability accuracy indicators, and feed them back to the prediction module, the reliability dynamic evaluation module, and the vehicle dispatching module for parameter optimization. Specifically, the feedback loop includes: retraining the time series prediction model using demand error; updating the classification model incrementally using Brier scores and adjusting dynamic factor weights; and adaptively adjusting based on demand fulfillment rate and peak occurrence rate. Weights.

[0053] The above modules work together to achieve high-precision prediction and low-cost scheduling of vehicle supply and demand in the Internet of Vehicles environment.

[0054] Example 3 Embodiment 3 of the present invention provides an electronic device.

[0055] An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor. The processor includes, but is not limited to, at least one of a central processing unit (CPU), a graphics processing unit (GPU), a neural network processor (NPU), a tensor processor (TPU), or an artificial intelligence acceleration chip. When executing the program, the processor implements the steps in the vehicle intelligent scheduling method based on the Internet of Vehicles as described in Embodiment 1 of the present invention.

[0056] The detailed steps are the same as those of the vehicle-to-everything (V2X) intelligent dispatching method provided in Example 1, and will not be repeated here.

[0057] Example 4 Embodiment 4 of the present invention provides a computer-readable storage medium.

[0058] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the vehicle intelligent scheduling method based on the Internet of Vehicles as described in Embodiment 1 of the present invention.

[0059] The detailed steps are the same as those of the vehicle-to-everything (V2X) intelligent dispatching method provided in Example 1, and will not be repeated here.

[0060] Example 5 Embodiment 5 of the present invention provides a computer program product.

[0061] A computer program product includes software code, wherein the program in the software code performs the steps of the vehicle intelligent scheduling method based on the Internet of Vehicles as described in Embodiment 1 of the present invention.

[0062] The detailed steps are the same as those of the vehicle-to-everything (V2X) intelligent dispatching method provided in Example 1, and will not be repeated here.

[0063] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages. For example, in one implementation, the methods and systems can be developed based on deep learning frameworks (such as TensorFlow, PyTorch, etc.) and using the Python language. Those skilled in the art will understand that other suitable programming languages ​​or tools can also be used for implementation without departing from the core ideas of the present invention.

[0064] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0065] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0066] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0067] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A vehicle intelligent scheduling method based on the Internet of Vehicles, characterized in that, include: Obtain the number of available vehicles at each location at the current time, the vehicle usage plans and real-time GPS trajectories of unfinished orders, the historical temporary vehicle pick-up demand sequence and external characteristics, and the known external characteristics for future periods; For each unfinished order, the reliability of returning the car as planned is dynamically assessed based on its usage plan and real-time GPS trajectory. The weighted planned return volume for each time period and each network point is calculated using this reliability as the weight. The historical temporary vehicle retrieval demand sequence and external features are input into a pre-trained time series prediction model to obtain the predicted values ​​of temporary vehicle retrieval demand for each location in the future time period. The predicted value of temporary car pick-up demand is input into the temporary car return conversion model based on the joint distribution of rental period and return location to obtain the predicted value of temporary car return volume for each location in the future time period. The total vehicle pick-up demand is obtained by adding the reserved vehicle pick-up demand to the predicted temporary vehicle pick-up demand, and the total vehicle return supply is obtained by adding the weighted planned vehicle return quantity to the predicted temporary vehicle return quantity. Then, the net demand of each outlet in the future for each time period is calculated. Based on the net demand, available vehicles, network capacity, and scheduling cost, a multi-objective scheduling optimization model is constructed and solved to obtain the number of vehicles dispatched from each network point to each network point and the number of dispatches, which are then output as scheduling instructions.

2. The method according to claim 1, characterized in that, The dynamic assessment of the reliability of returning each incomplete order to the vehicle as scheduled specifically includes: Obtain basic order characteristics, user profile characteristics, and static environmental characteristics at the time of order generation; The above features are input into a pre-trained classification model to obtain the initial credibility score of the order; During vehicle use, the real-time GPS trajectory sequence of the vehicle is acquired, and the direction consistency factor, abnormal stop factor and path deviation factor are extracted. The factors are then time-weighted and modulated according to the vehicle use progress to obtain dynamic adjustment coefficients. The initial confidence score is multiplied by the dynamic adjustment coefficient, and then subjected to amplitude limiting to obtain the real-time confidence score at the current moment. The real-time reliability at the current moment is calculated using the following formula: ； in, For orders The initial credibility score, For the first A dynamic adjustment coefficient.

3. The method according to claim 2, characterized in that, The classification model includes any one or more of the following: decision tree-based ensemble learning model, logistic regression model, and support vector machine model; the time series prediction model includes any one or more of the following: long short-term memory network, gated recurrent unit, recurrent neural network, and support vector regression model.

4. The method according to claim 1, characterized in that, The calculation method for the predicted temporary vehicle return volume is as follows: Get the number of orders from each pick-up location, each rental period, and each return location from the historical order data; A joint probability table of rental period and return location was obtained, where each element represents the distance from the pick-up point to the drop-off point. Departure, rental period is During a specific time period, or at the branch probability ; For any future time period Temporary car retrieval demand forecast Multiply this by the corresponding joint probability distribution, and sum over all pick-up times and pick-up locations to obtain the probability for any future time period. At any point The predicted value of temporary car return volume; The predicted value of temporary vehicle return volume is calculated using the following formula: ； in, For from the outlet Dispatch a vehicle to the branch. The cost, This represents the total number of time periods to be predicted.

5. The method according to claim 1, characterized in that, The construction and solution of the multi-objective scheduling optimization model specifically includes: The objective function is to minimize the sum of the weighted scheduling mileage cost and the weighted scheduling count cost, where the mileage cost is weighted. Weighted by frequency cost satisfy ; Set constraints on vehicle flow conservation, available vehicles, network capacity, the relationship between scheduling frequency and vehicle number, state transition equations, and initial conditions. The optimization model is solved using a mixed-integer linear programming solver or a heuristic algorithm to obtain the network points for each time period. Dispatch to branch Number of vehicles and number of scheduling ; Only execute the scheduling instructions for the current time period, and solve the problem on a rolling basis at the next decision time; The specific form of the objective function is as follows: ； in, For from the outlet Dispatch a vehicle to the branch. The cost, This represents the total number of time periods to be predicted.

6. The method according to claim 2, characterized in that, It also includes closed-loop self-learning steps: Collect actual operational data, including actual vehicle pick-up demand, actual vehicle return volume, actual return time and location for each order, and execution results of dispatch instructions at each time period and location. Calculate the prediction error of temporary vehicle retrieval demand and feed the error back into the time series prediction model for retraining; Calculate the Brier score between the credibility of each order and whether the vehicle was actually returned as planned, and feed this score back to the classification model for incremental updates, while adjusting the weights of the direction consistency factor, abnormal stay factor, and path deviation factor. The mileage cost weight in the multi-objective scheduling optimization model is adaptively adjusted based on the actual demand satisfaction rate and the occurrence rate of peak traffic. Weighted by frequency cost .

7. A vehicle intelligent dispatching system based on the Internet of Vehicles, characterized in that, include: The data acquisition module is used to acquire the number of available vehicles at each network point at the current time, the vehicle usage plans and real-time GPS trajectories of unfinished orders, the historical temporary vehicle pick-up demand sequence and external characteristics, and the known external characteristics of future time periods; The credibility dynamic assessment module is used to dynamically assess the credibility of returning the car as planned for each incomplete order based on its car usage plan and real-time GPS trajectory, and calculate the weighted planned car return volume for each time period and each network point in the future using the credibility as the weight. The prediction module is used to input the historical temporary vehicle retrieval demand sequence and external features into a pre-trained time series prediction model to obtain the predicted values ​​of temporary vehicle retrieval demand for each location in the future time period. The temporary vehicle return statistics module is used to input the predicted value of temporary vehicle pick-up demand into the temporary vehicle return conversion model based on the joint distribution of rental period and return location to obtain the predicted value of temporary vehicle return volume for each location in the future. The supply and demand fusion calculation module is used to add the reserved car pick-up demand to the predicted value of the temporary car pick-up demand to obtain the total car pick-up demand, and add the weighted planned car return quantity to the predicted value of the temporary car return quantity to obtain the total car return supply, and then calculate the net demand of each network point in the future for each time period. The vehicle scheduling module is used to construct and solve a multi-objective scheduling optimization model based on the net demand, the number of available vehicles, the network capacity, and the scheduling cost, and to obtain the number of vehicles dispatched from each network to each network and the number of scheduling operations, which are then output as scheduling instructions.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the steps of the vehicle intelligent scheduling method based on the Internet of Vehicles according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the vehicle intelligent scheduling method based on the Internet of Vehicles as described in any one of claims 1 to 6.

10. A computer program product, comprising software code, characterized in that, The program in the software code executes the steps of the vehicle intelligent scheduling method based on the Internet of Vehicles according to any one of claims 1 to 6.

Images