Online car-hailing platform scheduling system and method based on data collaboration, and storage medium
By using a data-driven ride-hailing platform dispatch system that combines intelligent dispatch, revenue distribution, and load forecasting modules, the system addresses the shortcomings of existing platforms in terms of unified modeling logic and interactive feedback mechanisms. This enables efficient resource scheduling and power load forecasting, thereby improving user experience and grid management capabilities.
Patent Information
- Application Number
- CN202511112076.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-08
- Publication Date
- 2025-11-21
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing ride-hailing platforms lack a unified modeling logic and interactive feedback mechanism among capacity scheduling, power distribution, and commercial revenue, resulting in low resource utilization, discontinuous user charging experience, difficulty in accurately predicting grid load, and difficulty in responding to the rapid response requirements of fast charging technology.
A data-driven ride-hailing platform dispatch system is adopted, which combines an intelligent dispatch module, a revenue distribution module, and a load forecasting module. By utilizing Transformer encoders, reinforcement learning, graph neural networks, and Shapley value theory, dynamic resource scheduling, fair revenue distribution, and high-resolution power load forecasting are achieved.
It has realized a global scheduling mechanism for ride-hailing platforms, improved resource utilization, optimized driver income distribution, improved the accuracy of power load forecasting, solved the problems of insufficient coordination capabilities and grid perception lag in existing platforms, and met the rapid response requirements of megawatt-level ultra-fast charging.
Smart Images
Figure CN120996461A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of smart energy technology, and in particular to a data-collaborative ride-hailing platform scheduling system, method, and storage medium. Background Technology
[0002] With the rapid increase in the penetration rate of electric vehicles in cities, ride-hailing services have become an important carrier of new energy consumption, posing greater challenges to the existing public charging network in terms of both quantity and timeliness of their electricity replenishment needs. Currently, as a high-frequency, high-mileage mode of transportation, ride-hailing services have significantly higher requirements for charging efficiency, network coverage, and dispatch response than private vehicles. The existing public charging network is gradually becoming inadequate to meet the characteristics of this scenario in terms of resource allocation flexibility and the accuracy of supply-demand matching. At the same time, the concentration of charging activities (such as during peak hours) and uneven regional distribution further exacerbate the supply-demand imbalance of energy replenishment resources, putting all parties in the industry chain (ride-hailing platforms, charging operators, power grid companies, etc.) under dual pressure of collaborative efficiency and operating costs.
[0003] The gradual deployment of megawatt-level ultra-fast charging technology has become a significant technological trend in the industry. This technology, by drastically reducing charging time (down to less than 15 minutes), effectively meets the core needs of ride-hailing services for rapid energy replenishment and efficient operation, and is expected to reshape the service model of public charging networks. However, the implementation of ultra-fast charging technology not only involves upgrading charging equipment but also poses new requirements for grid load capacity, power distribution strategies for charging stations, and coordination with real-time dispatching of ride-hailing vehicles.
[0004] These problems are mainly reflected in the following aspects:
[0005] Rigid scheduling mechanism: The existing charging scheduling is dominated by the platform side and relies on fixed strategies or experience rules. It lacks the ability to dynamically respond to time-varying conditions (such as real-time traffic flow and charging pile failures), regional heterogeneous conditions (differences in charging demand in different regions), and resource competition, resulting in low resource utilization.
[0006] Insufficient collaboration capabilities: Ride-hailing platforms and third-party charging station operators mostly adopt a static affiliation model, resulting in a lack of data interoperability and deep scheduling collaboration. This leads to problems such as discontinuous user charging experience (e.g., mismatch between reserved and actually available charging stations) and redundant platform operations (high cost of empty driving to find charging stations).
[0007] Lagging power grid perception: The load side of the power grid has difficulty sensing regional energy consumption trends in real time through platform behavior, and cannot accurately predict charging load fluctuations. This exacerbates the difficulty of coupling urban power allocation and supply and demand response, and may lead to the risk of local power grid overload.
[0008] In summary, existing industry platforms lack a system-level solution that can establish a unified modeling logic and interactive feedback mechanism among capacity scheduling, power distribution, and commercial revenue. Even if some platforms introduce basic prediction mechanisms, their modeling granularity and response chain are far below the rapid response requirements of megawatt-level ultra-fast charging technology, and they cannot cope with the interactive effects of multiple factors such as regional user behavior differences, traffic bottleneck migration, and electricity price fluctuations. Summary of the Invention
[0009] This invention aims to solve the technical problem that existing platforms lack a unified modeling logic and interactive feedback mechanism among capacity scheduling, power distribution, and commercial revenue.
[0010] To address the aforementioned technical problems, in a first aspect, the present invention provides a data-collaboration-based ride-hailing platform dispatch system, comprising:
[0011] The intelligent scheduling module is used to acquire ride-hailing vehicle operation data and data on the ride-hailing vehicle operation area including charging resources, and to perform scheduling analysis on the ride-hailing vehicle data and the ride-hailing vehicle operation area data through an intelligent scheduling algorithm based on fusion attention mechanism and reinforcement learning, so as to obtain an updated scheduling strategy for ride-hailing vehicles and charging resources.
[0012] The revenue distribution module is used to obtain a revenue distribution plan for ride-hailing drivers based on the ride-hailing operation data and the ride-hailing operation area data, according to a preset behavior adjustment method.
[0013] The load prediction module is used to construct a regional heterogeneous map based on the data of the ride-hailing operation area, and to perform inference propagation of the regional heterogeneous map through a preset graph neural network to obtain the power load prediction result.
[0014] Furthermore, the intelligent scheduling algorithm in the intelligent scheduling module is specifically as follows:
[0015] Within each scheduling cycle, the original vectors of the ride-hailing operation data and the ride-hailing operation area data are converted into context embedding vectors using a Transformer encoder;
[0016] The coupling relationship between ride-hailing behavior and charging resources is obtained based on the context embedding vector, and the dispatching actions of ride-hailing drivers are obtained.
[0017] Calculate the real-time return value for ride-hailing drivers regarding order completion efficiency, waiting time, and electricity costs based on the dispatch actions;
[0018] Based on the aforementioned reward value, the original scheduling strategy of the ride-hailing platform is updated using the standard Bellman equation to obtain the updated scheduling strategy.
[0019] Furthermore, the context embedding vector is defined as s. t The scheduling action is It satisfies the following relationship:
[0020] π φ (a t |s t ) = softmax(Qφ(s) t ,a t ));
[0021] Where, π φ (a t |s t ) represents a state s given at time t. t Take action a at the time t The probability, Qφ(s) t ,a t ) is the action value function, and softmax is the probability distribution function;
[0022] The instant reward value is defined as r. i It satisfies the following relationship:
[0023] r i =λ1·Order completion rate -λ2·Queue time -λ3·Idle time;
[0024] Wherein, λ1, λ2, and λ3 are weight values;
[0025] The process of updating the original scheduling strategy of a ride-hailing platform using the standard Bellman equation satisfies the following relationship:
[0026]
[0027] Where a is the learning rate and γ is the future return discount factor.
[0028] Furthermore, the preset behavior adjustment method in the revenue distribution module is specifically as follows:
[0029] Participants are identified from the ride-hailing operation data and the ride-hailing operation area data, and the average marginal contribution of each participant is represented by the Shapley value.
[0030] The service completion rate, online time ratio, and response success rate of ride-hailing drivers within a scheduling cycle are used as indicators to construct a behavior control factor, and the behavior control factor is introduced into the Shapley value to obtain a modified Shapley value.
[0031] The revenue distribution scheme for ride-hailing drivers within the current scheduling period is calculated based on the modified Shapley value.
[0032] Furthermore, the Shapley value is defined as φ. i It satisfies the following relationship:
[0033]
[0034] Where i represents a participant, N is the entire set of participants, v(S) represents the total benefit brought about by the cooperation of its subset S, and φ i This represents the average marginal contribution of each participating individual.
[0035] The regulating factor is defined as θ i It satisfies the following relationship:
[0036] θ i =a1v i +a2t i +a3p i ;
[0037] Among them, v i To achieve service completion rate, t i p represents the percentage of online time. i To improve the success rate of the response;
[0038] The modified Shapley value is defined as φ. i ′, which satisfies the following relationship:
[0039]
[0040] Where, θ i The adjustment factor is j, which is the subscript used to represent the participating entity.
[0041] Furthermore, the load prediction module is specifically used for:
[0042] Based on the ride-hailing operation area data, each area is treated as a node, and the physical connections between areas are treated as edges to construct the regional heterogeneous graph. The ride-hailing resource characteristics of each node are represented as a time series feature vector.
[0043] The heterogeneous regional graph is inferred and propagated through the preset graph neural network to obtain the power load prediction result.
[0044] Furthermore, the goal of the preset graph neural network is to minimize the weighted mean square error between the predicted value and the true value.
[0045] Secondly, the present invention also provides a data-collaborative ride-hailing platform scheduling method, comprising the following steps:
[0046] The system acquires ride-hailing vehicle operation data and data on the ride-hailing vehicle operation area containing charging resources. Then, it performs scheduling analysis on the ride-hailing vehicle data and the ride-hailing vehicle operation area data using an intelligent scheduling algorithm based on a fusion attention mechanism and reinforcement learning, thereby obtaining an updated scheduling strategy for ride-hailing vehicles and the charging resources.
[0047] Based on the ride-hailing operation data and the ride-hailing operation area data, a revenue distribution plan for ride-hailing drivers is obtained according to a preset behavior adjustment method;
[0048] Based on the data of the ride-hailing operation area, a regional heterogeneous map is constructed, and the regional heterogeneous map is inferred and propagated through a preset graph neural network to obtain the power load prediction result.
[0049] Thirdly, the present invention also provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the data-cooperative ride-hailing platform scheduling method as described in any of the above embodiments.
[0050] Fourthly, the present invention also provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the data-cooperative ride-hailing platform scheduling method as described in any of the above embodiments.
[0051] The beneficial effect achieved by this invention lies in proposing a data-collaborative ride-hailing platform scheduling system. This system conducts data collaboration and algorithm interconnection from three perspectives: resource scheduling, driver income, and resource perception. It also combines reinforcement learning, game theory contribution theory, and graph neural network inference to realize a global scheduling mechanism, thus making up for the shortcomings of existing platforms. Attached Figure Description
[0052] Figure 1 This is a schematic diagram of the structure of the ride-hailing platform dispatch system based on data collaboration provided in this invention.
[0053] Figure 2 This is a schematic diagram of the execution logic of the intelligent scheduling module provided in this invention.
[0054] Figure 3 This is a schematic diagram of the execution logic of the revenue distribution module provided in this invention.
[0055] Figure 4 This is a schematic diagram of the execution logic of the load prediction module provided in this invention.
[0056] Figure 5 This is a flowchart illustrating the steps of the ride-hailing platform scheduling method based on data collaboration provided in this invention.
[0057] Figure 6 This is a schematic diagram of the structure of the computer device provided in the embodiment of the present invention. Detailed Implementation
[0058] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0059] Please refer to Figure 1 , Figure 1 This is a schematic diagram of the structure of a data-collaboration-based ride-hailing platform dispatch system 100 provided in this invention. The data-collaboration-based ride-hailing platform dispatch system 100 includes:
[0060] The intelligent scheduling module 101 is used to acquire ride-hailing operation data and data on the ride-hailing operation area including charging resources, and to perform scheduling analysis on the ride-hailing data and the ride-hailing operation area data through an intelligent scheduling algorithm based on fusion attention mechanism and reinforcement learning, so as to obtain an updated scheduling strategy for ride-hailing vehicles and charging resources.
[0061] The revenue distribution module 102 is used to obtain a revenue distribution plan for ride-hailing drivers based on the ride-hailing operation data and the ride-hailing operation area data, according to a preset behavior adjustment method.
[0062] The load prediction module 103 is used to construct a regional heterogeneous map based on the data of the ride-hailing operation area, and to perform inference propagation of the regional heterogeneous map through a preset graph neural network to obtain the power load prediction result.
[0063] Specifically, the ride-hailing system employs a three-layer cooperation model, where operators, platforms, and power companies act as data and benefit providers and demanders, respectively. Based on this layered model, this invention proposes a Joint Attention & Reinforcement Dispatching Algorithm (JARDA), which integrates joint attention and reinforcement learning. This algorithm aims to dynamically and intelligently allocate drivers and charging resources on the platform, constructing a deep scheduling strategy network based on multi-source states. Specifically, the intelligent scheduling algorithm in the intelligent scheduling module 101 is as follows:
[0064] Within each scheduling cycle, the original vectors of the ride-hailing operation data and the ride-hailing operation area data are converted into context embedding vectors through a Transformer encoder. This encoding method can preserve global attention information, enabling each scheduling unit (driver, charging pile, area) to perceive the resource pressure and behavioral coupling relationship in the entire city system.
[0065] The coupling relationship between ride-hailing behavior and charging resources is obtained based on the context embedding vector, and the dispatching actions of ride-hailing drivers are obtained. In this embodiment of the invention, the dispatching actions include going to different charging piles to charge, continuing to execute order tasks, waiting in place, and migrating across regions.
[0066] Calculate the real-time return value for ride-hailing drivers regarding order completion efficiency, waiting time, and electricity costs based on the dispatch actions;
[0067] Based on the aforementioned reward value, the original scheduling strategy of the ride-hailing platform is updated using the standard Bellman equation to obtain the updated scheduling strategy.
[0068] Furthermore, the context embedding vector is defined as s. t The scheduling action is It satisfies the following relationship:
[0069] π φ (a t |s t ) = softmax(Qφ(s) t ,a t ));
[0070] Wherein, π φ (a t |s t ) represents a state s given at time t. t Take action a at the time t The probability, Qφ(s) t ,a t ) is the action value function proposed in the embodiments of the present invention, and softmax is the probability distribution function.
[0071] The instant reward value is defined as r. i It satisfies the following relationship:
[0072] r i =λ1·Order completion rate -λ2·Queue time -λ3·Idle time;
[0073] Wherein, λ1, λ2, and λ3 are weight values;
[0074] The process of updating the original scheduling strategy of a ride-hailing platform using the standard Bellman equation satisfies the following relationship:
[0075]
[0076] Where a is the learning rate and γ is the future return discount factor.
[0077] like Figure 2 The diagram illustrates the execution logic of the intelligent scheduling module 101. The core idea of the intelligent scheduling algorithm lies in transforming the traditional scheduling problem from rule matching to policy optimization. Through the global attention mechanism of the Transformer encoder, the modeling in this embodiment can perceive the coupling relationships between any areas in the city, such as energy transfer paths between hot and non-hot areas. Furthermore, the reinforcement learning method based on the standard Bellman equation allows the model to autonomously evolve its policy according to long-term goals (such as maximizing service rate and improving platform revenue), thus eliminating the dependence of existing algorithms on expert rules.
[0078] Based on the design of the intelligent scheduling module 101, to further improve the cooperation efficiency and system fairness among the platform participants, this embodiment of the invention introduces an Adjustable Shapley Revenue Allocation (A-Shapley) mechanism. This mechanism constructs a basic framework for platform revenue allocation based on Shapley value theory in game theory, and introduces behavioral adjustment factors to incorporate the service performance of each participant (such as drivers, charging pile operators, and the platform itself) in a specific period into revenue adjustment. Specifically, the preset behavioral adjustment method in the revenue allocation module 102 is as follows:
[0079] Participants are identified from the ride-hailing operation data and the ride-hailing operation area data, and the average marginal contribution of each participant is represented by the Shapley value.
[0080] The service completion rate, online time ratio, and response success rate of ride-hailing drivers within a scheduling cycle are used as indicators to construct a behavior control factor, and the behavior control factor is introduced into the Shapley value to obtain a modified Shapley value.
[0081] The revenue distribution scheme for ride-hailing drivers within the current scheduling period is calculated based on the modified Shapley value.
[0082] Furthermore, the Shapley value is defined as φ. i It satisfies the following relationship:
[0083]
[0084] Where i represents a participant, N is the entire set of participants, v(S) represents the total benefit brought about by the cooperation of its subset S, and φ i This represents the average marginal contribution of each participating individual.
[0085] The regulating factor is defined as θ i It satisfies the following relationship:
[0086] θ i =a1v i +a2t i +a3p i ;
[0087] Among them, v i To achieve service completion rate, t i p represents the percentage of online time. i To improve the success rate of the response;
[0088] The modified Shapley value is defined as φ. i ′, which satisfies the following relationship:
[0089]
[0090] Where, θ i The adjustment factor is j, which is the subscript used to represent the participating entity.
[0091] like Figure 3 The execution logic diagram of the revenue distribution module 102 shown below preserves the fairness and game consistency of the original distribution mechanism by modifying the Shapley value, while guiding participants towards more efficient behavior in actual operation. For example, with the same contribution, drivers who respond more actively to orders or stay online for longer will receive more platform revenue; operators who prioritize providing charging station resources during peak electricity price periods or queuing relief periods will receive a higher return coefficient.
[0092] To further enhance the predictive capability between the dispatching system and the power grid supply and demand, this embodiment of the invention designs a regional power load prediction model (GNN-ECP) based on a graph neural network (GNN) as the load prediction module 103. This enables high-resolution regional-level power demand prediction and early warning, providing prior input for dispatching decisions and electricity pricing strategies. The load prediction module 103 is specifically used for:
[0093] Based on the ride-hailing operation area data, each area is treated as a node, and the physical connections between areas (such as road traffic, shared order routes, and power grid topology) are used as edges to construct the regional heterogeneous graph. The ride-hailing resource characteristics of each node are represented as a time series feature vector.
[0094] The heterogeneous regional graph is inferred and propagated through the preset graph neural network to obtain the power load prediction result.
[0095] During implementation, the preset graph neural network uses a graph convolutional network (GCN) or attention graph network (GAT) as the core propagation module. It uses node features such as meteorological data and historical load data within the region as prediction data sources for inference propagation, so as to support the modeling of highly heterogeneous correlations between different regions.
[0096] Furthermore, the objective of the preset graph neural network is to minimize the weighted mean square error between the predicted and true values. This objective can be expressed as:
[0097]
[0098] Where N represents the total number of samples used to calculate the loss (i.e., the total number of regions), w i This is used as a regional importance weight to highlight high-precision predictions for key areas (such as hospitals and transportation hubs). This represents the true value of the i-th region at time t+1.
[0099] like Figure 4 The diagram illustrates the execution logic of the load forecasting module 103. This module not only improves the spatial resolution of power load but also enhances trend perception capabilities over time. During implementation, by interfacing with the intelligent scheduling module 101, the platform can predict and implement control measures before power price increases, power shortages, or peak charging periods, such as guiding vehicle charging in advance or restricting order dispatch. Simultaneously, the regional power load forecast value output by the load forecasting module 103 can also serve as a constraint, feeding back the distribution weights affecting the Shapley value, thereby enabling the revenue distribution module 102 to jointly regulate supply and demand and revenue distribution.
[0100] The beneficial effect achieved by this invention lies in proposing a data-collaborative ride-hailing platform scheduling system. This system conducts data collaboration and algorithm interconnection from three perspectives: resource scheduling, driver income, and resource perception. It also combines reinforcement learning, game theory contribution theory, and graph neural network inference to realize a global scheduling mechanism, thus making up for the shortcomings of existing platforms.
[0101] This invention also provides a data-collaboration-based ride-hailing platform scheduling method, please refer to... Figure 5 , Figure 5 This is a flowchart illustrating the steps of a data-collaboration-based ride-hailing platform scheduling method provided in an embodiment of the present invention. The data-collaboration-based ride-hailing platform scheduling method includes the following steps:
[0102] S201. Obtain ride-hailing vehicle operation data and ride-hailing vehicle operation area data including charging resources, and perform scheduling analysis on the ride-hailing vehicle data and the ride-hailing vehicle operation area data through an intelligent scheduling algorithm based on fusion attention mechanism and reinforcement learning to obtain an updated scheduling strategy for ride-hailing vehicles and charging resources.
[0103] S202. Based on the ride-hailing operation data and the ride-hailing operation area data, obtain a revenue distribution plan for ride-hailing drivers according to a preset behavior adjustment method;
[0104] S203. Construct a regional heterogeneous map based on the ride-hailing operation area data, and perform inference propagation of the regional heterogeneous map through a preset graph neural network to obtain the power load prediction result.
[0105] The data-collaboration-based ride-hailing platform scheduling method described above can realize the relevant functions of the data-collaboration-based ride-hailing platform scheduling system as described in the above embodiments, and can achieve the same technical effect. Referring to the description in the above embodiments, it will not be repeated here.
[0106] This invention also provides a computer device, please refer to... Figure 6 , Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. The computer device 300 includes: a memory 302, a processor 301, and a computer program stored in the memory 302 and executable on the processor 301.
[0107] The processor 301 calls the computer program stored in the memory 302 to execute the steps in the method provided in this embodiment of the invention. Please refer to... Figure 5 Specifically, it includes the following steps:
[0108] S201. Obtain ride-hailing vehicle operation data and ride-hailing vehicle operation area data including charging resources, and perform scheduling analysis on the ride-hailing vehicle data and the ride-hailing vehicle operation area data through an intelligent scheduling algorithm based on fusion attention mechanism and reinforcement learning to obtain an updated scheduling strategy for ride-hailing vehicles and charging resources.
[0109] S202. Based on the ride-hailing operation data and the ride-hailing operation area data, obtain a revenue distribution plan for ride-hailing drivers according to a preset behavior adjustment method;
[0110] S203. Construct a regional heterogeneous map based on the ride-hailing operation area data, and perform inference propagation of the regional heterogeneous map through a preset graph neural network to obtain the power load prediction result.
[0111] The computer device 300 provided in this embodiment of the invention can implement the steps in the data-cooperative online ride-hailing platform scheduling method as described in the above embodiments, and can achieve the same technical effect. Referring to the description in the above embodiments, it will not be repeated here.
[0112] This invention also provides a storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes and steps of the data-cooperative ride-hailing platform scheduling method provided in this invention and achieves the same technical effect. To avoid repetition, it will not be described again here.
[0113] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by computer programs or related hardware (such as mobile phones, computers, servers, air conditioners, or network devices). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.
[0114] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
[0115] The embodiments of the present invention have been described above with reference to the accompanying drawings. The disclosed embodiments are merely preferred embodiments of the present invention. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many equivalent changes in form under the guidance of the present invention without departing from the spirit and scope of the claims. All such changes are within the protection scope of the present invention.
Claims
1. A data-collaborative online ride-hailing platform dispatch system, characterized in that, include: The intelligent scheduling module is used to acquire ride-hailing vehicle operation data and data on the ride-hailing vehicle operation area including charging resources, and to perform scheduling analysis on the ride-hailing vehicle data and the ride-hailing vehicle operation area data through an intelligent scheduling algorithm based on fusion attention mechanism and reinforcement learning, so as to obtain an updated scheduling strategy for ride-hailing vehicles and charging resources. The revenue distribution module is used to obtain a revenue distribution plan for ride-hailing drivers based on the ride-hailing operation data and the ride-hailing operation area data, according to a preset behavior adjustment method. The load prediction module is used to construct a regional heterogeneous map based on the data of the ride-hailing operation area, and to perform inference propagation of the regional heterogeneous map through a preset graph neural network to obtain the power load prediction result.
2. The ride-hailing platform dispatch system based on data collaboration according to claim 1, characterized in that, The intelligent scheduling algorithm in the intelligent scheduling module is specifically as follows: Within each scheduling cycle, the original vectors of the ride-hailing operation data and the ride-hailing operation area data are converted into context embedding vectors using a Transformer encoder; The coupling relationship between ride-hailing behavior and charging resources is obtained based on the context embedding vector, and the dispatching actions of ride-hailing drivers are obtained. Calculate the real-time return value for ride-hailing drivers regarding order completion efficiency, waiting time, and electricity costs based on the dispatch actions; Based on the aforementioned reward value, the original scheduling strategy of the ride-hailing platform is updated using the standard Bellman equation to obtain the updated scheduling strategy.
3. The ride-hailing platform dispatch system based on data collaboration according to claim 2, characterized in that, Define the context embedding vector as s t The scheduling action is It satisfies the following relationship: p φ (a t |s t )=softmax(Qφ(s t ,a t )); Where, π φ (a t |s t ) represents a state s given at time t. t Take action a at the time t The probability, Qφ(s) t ,a t ) is the action value function, and softmax is the probability distribution function; The instant reward value is defined as r. i It satisfies the following relationship: r i =λ1·Order completion rate -λ2·Queue time -λ3·Idle time; Wherein, λ1, λ2, and λ3 are weight values; The process of updating the original scheduling strategy of a ride-hailing platform using the standard Bellman equation satisfies the following relationship: Where a is the learning rate and γ is the future return discount factor.
4. The ride-hailing platform dispatch system based on data collaboration according to claim 1, characterized in that, The preset behavior adjustment method in the revenue distribution module is specifically as follows: Participants are identified from the ride-hailing operation data and the ride-hailing operation area data, and the average marginal contribution of each participant is represented by the Shapley value. The service completion rate, online time ratio, and response success rate of ride-hailing drivers within a scheduling cycle are used as indicators to construct a behavior control factor, and the behavior control factor is introduced into the Shapley value to obtain a modified Shapley value. The revenue distribution scheme for ride-hailing drivers within the current scheduling period is calculated based on the modified Shapley value.
5. The ride-hailing platform dispatch system based on data collaboration according to claim 4, characterized in that, The Shapley value is defined as φ. i It satisfies the following relationship: Where i represents a participant, N is the entire set of participants, v(S) represents the total benefit brought about by the cooperation of its subset S, and φ i This represents the average marginal contribution of each participating individual. The regulating factor is defined as θ i It satisfies the following relationship: θ i <a1v i +a2t i +a3p i ; Among them, v i To achieve service completion rate, t i p represents the percentage of online time. i To improve the success rate of the response; The modified Shapley value is defined as φ. i ′, which satisfies the following relationship: Where, θ i The adjustment factor is j, which is the subscript used to represent the participating entity.
6. The ride-hailing platform dispatch system based on data collaboration according to claim 1, characterized in that, The load prediction module is specifically used for: Based on the ride-hailing operation area data, each area is treated as a node, and the physical connections between areas are treated as edges to construct the regional heterogeneous graph. The ride-hailing resource characteristics of each node are represented as a time series feature vector. The heterogeneous regional graph is inferred and propagated through the preset graph neural network to obtain the power load prediction result.
7. The ride-hailing platform dispatch system based on data collaboration according to claim 6, characterized in that, The goal of the preset graph neural network is to minimize the weighted mean square error between the predicted value and the true value.
8. A data-collaborative method for dispatching ride-hailing platforms, characterized in that, Includes the following steps: The system acquires ride-hailing vehicle operation data and data on the ride-hailing vehicle operation area containing charging resources. Then, it performs scheduling analysis on the ride-hailing vehicle data and the ride-hailing vehicle operation area data using an intelligent scheduling algorithm based on a fusion attention mechanism and reinforcement learning, thereby obtaining an updated scheduling strategy for ride-hailing vehicles and the charging resources. Based on the ride-hailing operation data and the ride-hailing operation area data, a revenue distribution plan for ride-hailing drivers is obtained according to a preset behavior adjustment method; Based on the data of the ride-hailing operation area, a regional heterogeneous map is constructed, and the regional heterogeneous map is inferred and propagated through a preset graph neural network to obtain the power load prediction result.
9. A computer device, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the data-cooperative ride-hailing platform scheduling method as described in claim 8.
10. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps in the data-cooperative ride-hailing platform scheduling method as described in claim 8.