Real-time data driven demand response bus intelligent scheduling method and system

Abstract

The invention provides a demand response bus intelligent scheduling method and system driven by real-time data. A demand response bus intelligent scheduling method driven by real-time data comprises the following steps: S1, collecting and fusing multi-source data in real time, and constructing a dynamic data set; s2, carrying out short-time demand prediction and clustering analysis based on LSTM and DBSCAN; s3, dynamic service area division and greedy algorithm line generation; s4, elastic vehicle scheduling based on combination of mixed integer programming and a simulated annealing algorithm; s5, dynamically adjusting the real-time ticket price driven by the supply and demand game model; s6, passenger personalized service matching is achieved through a collaborative filtering algorithm; s7, performing multi-target simulation verification on the digital twin platform; and S8, reinforcing a learning-driven dynamic feedback optimization mechanism. According to the real-time data-driven demand response bus intelligent scheduling method and system provided by the invention, the operation efficiency, the passenger satisfaction degree and the enterprise benefit of the demand response type customized bus can be remarkably improved, and the method and the system are suitable for construction of an urban intelligent bus system and have remarkable advantages.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent scheduling, and particularly to a real-time data-driven demand-responsive bus intelligent scheduling method and system. Background Art

[0002] With the acceleration of the urbanization process and the increasing diversification of residents' travel demands, the traditional fixed-route bus system faces severe challenges in terms of flexibility, efficiency, and passenger experience. As a new type of public transportation mode, Demand-Responsive Transit (DRT) can effectively make up for the deficiencies of conventional buses by dynamically adjusting routes, vehicle scheduling, and fare strategies, and has become an important part of the urban intelligent transportation system. However, there are still many bottlenecks in the technical implementation of existing DRT systems, which restrict their large-scale promotion and application.

[0003] Therefore, it is necessary to provide a real-time data-driven demand-responsive bus intelligent scheduling method and system to solve the above technical problems. Summary of the Invention

[0004] The technical problem solved by the present invention is to provide a real-time data-driven demand-responsive bus intelligent scheduling method and system that can significantly improve the operation efficiency, passenger satisfaction, and enterprise revenue of customized buses and is applicable to the construction of urban intelligent transportation systems.

[0005] To solve the above technical problems, the real-time data-driven demand-responsive bus intelligent scheduling method provided by the present invention includes the following steps:

[0006] S1: Real-time collection and fusion of multi-source data to construct a dynamic data set;

[0007] S2: Short-term demand prediction and clustering analysis based on LSTM and DBSCAN;

[0008] S3: Dynamic service area division and greedy algorithm route generation;

[0009] S4: Flexible vehicle scheduling combining mixed integer programming and simulated annealing algorithm;

[0010] S5: Real-time fare dynamic adjustment driven by a supply-demand game model;

[0011] S6: Passenger personalized service matching implemented by collaborative filtering algorithm;

[0012] S7: Multi-objective simulation verification of the digital twin platform;

[0013] S8: Dynamic feedback optimization mechanism driven by reinforcement learning.

[0014] Preferably, in S1, data on passenger travel demands, vehicle positions, road congestion indices, and weather conditions are collected in real time through in-vehicle GPS, passenger reservation APPs, traffic flow sensors, and weather monitoring system devices.

[0015] Preferably, federated learning technology is used to fuse decentralized data under privacy protection to construct a dynamic data set, and the formula is as follows:

[0016] D t ={D GPS , D APP , D Traffic , D Weather ,}

[0017] Among them, D t is the fused data set within the time window t, and each sub-data set corresponds to a different source.

[0018] Preferably, in S2, a short-term demand prediction model is constructed based on the LSTM neural network to predict the passenger travel hot spots and demand within the next 15 minutes. The DBSCAN algorithm is used to perform spatio-temporal clustering on demand points to identify high-density demand clusters, and the formula is as follows:

[0019] Cluster(p i )=arg minC j ||P i -μ j ||≤∈

[0020] Among them, p i is the demand point, μ j is the clustering center, and ∈ is the neighborhood radius threshold.

[0021] Preferably, in S3, according to the demand cluster distribution and real-time road conditions, the Voronoi diagram is used to divide the dynamic service area. The greedy algorithm is combined to generate the initial route, which preferentially covers high-density demand clusters while restricting the route length not to exceed the preset threshold L max , ensuring that the operating cost is controllable.

[0022] Preferably, in S4, based on the mixed integer programming model, with the dual objectives of minimizing the operating cost (fuel, time) and maximizing the passenger load factor, vehicles are dynamically allocated to each route, and the simulated annealing algorithm is introduced to optimize the route, avoiding congested sections in real time. The update formula is as follows:

[0023] min(αΣ k C K +β∑ i,j T ij )

[0024] Among them, C k is the operating cost of vehicle k, Tij is the travel time of section i→j, and α, β are weight coefficients.

[0025] Preferably, in S5, according to demand elasticity, vehicle full load rate and time period characteristics, a game theory model is used to balance enterprise revenue and passenger willingness to pay, and the fare F is dynamically adjusted according to the real-time supply-demand ratio R:

[0026]

[0027] where R eq is the supply-demand balance threshold, and γ is the adjustment coefficient.

[0028] Preferably, in S6, based on passengers' historical travel data and real-time preferences, a collaborative filtering algorithm is used to recommend the optimal route, and a personalized ride plan is pushed through the APP to improve satisfaction.

[0029] The present invention also provides a real-time data-driven demand response bus intelligent scheduling system, including the following modules:

[0030] A multi-source data collection and fusion module, which is used to collect and fuse multi-source heterogeneous data in real time;

[0031] A demand prediction and clustering module, which realizes short-term demand prediction and clustering based on LSTM and DBSCAN algorithms;

[0032] A dynamic programming and scheduling module, which uses the Voronoi diagram and greedy algorithm to generate dynamic routes;

[0033] A vehicle scheduling optimization module, which optimizes the path through mixed integer programming and simulated annealing algorithms;

[0034] A fare dynamic management module, which adjusts the real-time fare based on the supply-demand game model;

[0035] A user service matching module, which pushes personalized solutions through a collaborative filtering algorithm;

[0036] A simulation verification module, which verifies the scheduling plan on the digital twin platform;

[0037] A feedback optimization module, which realizes closed-loop parameter optimization based on reinforcement learning.

[0038] Preferably, the feedback optimization module receives the inputs of the simulation verification module and the user service matching module, dynamically adjusts the scheduling parameters and feeds them back to the dynamic programming and scheduling module.

[0039] Compared with the related technologies, the real-time data-driven demand response bus intelligent scheduling method and system provided by the present invention have the following beneficial effects:

[0040] The present invention provides a real-time data-driven demand response bus intelligent scheduling method and system, which realizes accurate demand prediction by fusing multi-source data, optimizes route and vehicle scheduling by combining the greedy algorithm, mixed integer programming and simulated annealing algorithm, introduces a dynamic fare strategy to balance the supply and demand relationship, and uses a digital twin platform and reinforcement learning to achieve closed-loop optimization, which can significantly improve the operation efficiency, passenger satisfaction and enterprise benefits of demand response customized buses, and is applicable to the construction of urban intelligent bus systems. Description of the Drawings

[0041] Figure 1 is a flowchart of the real-time data-driven demand response bus intelligent scheduling method provided by the present invention;

[0042] Figure 2 is a schematic block diagram of the real-time data-driven demand response bus intelligent scheduling system provided by the present invention. Detailed Embodiments

[0043] The present invention will be further described below in conjunction with the drawings and embodiments.

[0044] Please refer to Figure 1 and Figure 2 , wherein, Figure 1 is a flowchart of the real-time data-driven demand response bus intelligent scheduling method provided by the present invention; Figure 2 is a schematic block diagram of the real-time data-driven demand response bus intelligent scheduling system provided by the present invention. The real-time data-driven demand response bus intelligent scheduling system includes: a multi-source data collection and fusion module: real-time data is collected through devices such as in-vehicle GPS, passenger APPs, traffic sensors, and meteorological systems, and federated learning technology is used for data fusion under privacy protection to construct a dynamic data set.

[0045] A demand prediction and clustering module: predicts short-term travel demand based on an LSTM neural network, and performs spatio-temporal clustering on demand points through the DBSCAN algorithm to generate high-density demand clusters.

[0046] A dynamic programming and scheduling module: uses a Voronoi diagram to divide the dynamic service area, combines the greedy algorithm to generate an initial route, and constrains the route length (L ≤ L max ).

[0047] A vehicle scheduling optimization module: based on a mixed integer programming (MIP) model and a simulated annealing algorithm, optimizes vehicle route allocation to minimize the operating cost (min(α∑ k C K +β∑ i,j T ij ))).

[0048] The fare dynamic management module adjusts the fare dynamically according to the supply-demand ratio (R) Balance the enterprise revenue and the willingness of passengers to pay.

[0049] User service matching module: Analyze the historical preferences of passengers based on the collaborative filtering algorithm, recommend personalized routes, and push real-time ride plans (including fares and estimated arrival times) through the APP.

[0050] Simulation verification module: Inject real-time data streams on the digital twin platform, and use the Monte Carlo method to verify the passenger capacity rate, on-time rate, and cost efficiency of the scheduling plan.

[0051] Feedback optimization module: Based on the reinforcement learning model, dynamically adjust the scheduling parameters (such as path weights and departure intervals) according to the simulation results and actual operation data (complaint rate, empty running rate).

[0052] The present invention also provides a real-time data-driven demand response bus intelligent scheduling method, including the following steps:

[0053] S1: Real-time collection and fusion of multi-source data to construct a dynamic data set;

[0054] Through in-vehicle GPS, passenger reservation APP, traffic flow sensors, and weather monitoring system devices, real-time collect data on passenger travel demands, vehicle positions, road congestion indexes, and weather conditions. Use federated learning technology to fuse the dispersed data under privacy protection to construct a dynamic data set, and the formula is as follows:

[0055] D t ={D GPS , D APP , D Traffic , D Weather ,}

[0056] Among them, D t is the fused data set within the time window t, and each sub-data set corresponds to a different source.

[0057] S2: Short-term demand prediction and clustering analysis based on LSTM and DBSCAN;

[0058] Based on the LSTM neural network, construct a short-term demand prediction model to predict the hot spots and demand volumes of passenger travel within the next 15 minutes. Use the DBSCAN algorithm to perform spatio-temporal clustering on the demand points to identify high-density demand clusters, and the formula is as follows:

[0059] Cluster(p i ) = arg minC j ||P i -μ j ||≤∈

[0060] Among them, p i is the demand point, μj is the clustering center, and ∈ is the threshold of the neighborhood radius.

[0061] S3: Dynamic service area division and greedy algorithm route generation;

[0062] According to the demand cluster distribution and real-time road conditions, use the Voronoi diagram to divide the dynamic service area. Combine the greedy algorithm to generate the initial route, giving priority to covering high-density demand clusters, and at the same time restricting the route length not to exceed the preset threshold L max to ensure that the operating cost is controllable.

[0063] S4: Flexible vehicle scheduling combining mixed integer programming and simulated annealing algorithm;

[0064] Based on the mixed integer programming model, with the dual objectives of minimizing the operating cost (fuel, time) and maximizing the passenger occupancy rate, dynamically allocate vehicles to each route, introduce the simulated annealing algorithm to optimize the path, and avoid congested sections in real time. The update formula is:

[0065] min(αΣ k C K +β∑ i,j T ij )

[0066] where C k is the operating cost of vehicle k, T ij is the travel time of section i→j, and α, β are weight coefficients.

[0067] S5: Real-time fare dynamic adjustment driven by the supply-demand game model;

[0068] According to the demand elasticity, vehicle occupancy rate and time period characteristics, use the game theory model to balance the enterprise revenue and the passenger's willingness to pay. The fare F is dynamically adjusted with the real-time supply-demand ratio R:

[0069]

[0070] where R eq is the supply-demand balance threshold, and γ is the adjustment coefficient.

[0071] S6: Passenger personalized service matching implemented by collaborative filtering algorithm;

[0072] Based on the passenger's historical travel data and real-time preferences (such as seat type, travel time window), use the collaborative filtering algorithm to recommend the optimal route, and push the personalized travel plan through the APP to improve the satisfaction.

[0073] S7: Multi-objective simulation verification of the digital twin platform;

[0074] Build a urban traffic simulation environment on the digital twin platform, inject real-time data streams, and verify the passenger capacity rate, on-time rate, and cost efficiency of the scheduling plan. Use the Monte Carlo method to evaluate the robustness under different scenarios.

[0075] S8: A dynamic feedback optimization mechanism driven by reinforcement learning.

[0076] Based on the actual operation data (such as the passenger complaint rate and the vehicle empty running rate), establish a reinforcement learning model, dynamically adjust the scheduling parameters (such as the departure interval and the path weight), and form a closed-loop optimization mechanism of "prediction - execution - feedback".

[0077] Compared with related technologies, the real-time data-driven demand-responsive bus intelligent scheduling method and system provided by the present invention have the following beneficial effects:

[0078] The present invention provides a real-time data-driven demand-responsive bus intelligent scheduling method and system, which realizes accurate demand prediction by fusing multi-source data, optimizes route and vehicle scheduling by combining the greedy algorithm, mixed integer programming and simulated annealing algorithm, introduces a dynamic fare strategy to balance the supply and demand relationship, and uses the digital twin platform and reinforcement learning to achieve closed-loop optimization, which can significantly improve the operation efficiency, passenger satisfaction and enterprise revenue of customized buses, and is applicable to the construction of urban intelligent transportation systems.

[0079] The above are only the embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be similarly included in the patent protection scope of the present invention.

Claims

1. A real-time data-driven demand response bus intelligent scheduling method, characterized in that, It includes the following steps: S1: Real-time collection and fusion of multi-source data to construct a dynamic data set; S2: Short-term demand prediction and clustering analysis based on LSTM and DBSCAN; S3: Dynamic service area division and generation of greedy algorithm routes; S4: Flexible vehicle scheduling combining mixed integer programming and simulated annealing algorithm; S5: Real-time fare dynamic adjustment driven by the supply-demand game model; S6: Passenger personalized service matching implemented by collaborative filtering algorithm; S7: Multi-objective simulation verification of the digital twin platform; S8: Dynamic feedback optimization mechanism driven by reinforcement learning.

2. The real-time data-driven demand response bus intelligent scheduling method according to claim 1, wherein, In S1, through in-vehicle GPS, passenger reservation APP, traffic flow sensors, and weather monitoring system devices, data on passenger travel demand, vehicle location, road congestion index, and weather conditions are collected in real time.

3. The real-time data-driven demand response bus intelligent scheduling method according to claim 2, wherein Federated learning technology is used to fuse the dispersed data under privacy protection to construct a dynamic data set, and the formula is as follows: D t = {D GPS , D APP , D Traffic , D Weather ,} Among them, D t is the fused data set within the time window t, and each sub-data set corresponds to a different source respectively.

4. The real-time data-driven demand response bus intelligent scheduling method according to claim 1, characterized in that In S2, a short-term demand prediction model is constructed based on the LSTM neural network to predict the passenger travel hot spots and demand within the next 15 minutes. The DBSCAN algorithm is used to perform spatio-temporal clustering on the demand points to identify high-density demand clusters, and the formula is as follows: Cluster(p i ) = argminC j ||P i - μ j || ≤ ∈ where p i is the demand point, μ j is the clustering center, and ∈ is the neighborhood radius threshold.

5. The real-time data-driven demand response bus intelligent scheduling method according to claim 1, characterized in that, In S3, according to the demand cluster distribution and real-time road conditions, a dynamic service area is divided using the Voronoi diagram. Combining with the greedy algorithm to generate an initial route, which preferentially covers high-density demand clusters while restricting the route length not to exceed the preset threshold L max , ensuring that the operating costs are controllable.

6. The real-time data-driven demand response bus intelligent scheduling method according to claim 1, characterized in that In S4, based on the mixed integer programming model, with the dual objectives of minimizing operating costs (fuel, time) and maximizing the passenger load factor, vehicles are dynamically allocated to each route. The simulated annealing algorithm is introduced to optimize the route and avoid congested sections in real time. The update formula is: min(αΣ k C K +β∑ i,j T ij ) Among them, C k is the operating cost of vehicle k, T ij is the travel time of section i→j, and α, β are weight coefficients.

7. The real-time data-driven demand response bus intelligent scheduling method according to claim 1, characterized in that In S5, according to demand elasticity, vehicle full load rate, and time period characteristics, a game theory model is used to balance the enterprise revenue and the passenger's willingness to pay. The fare F is dynamically adjusted with the real-time supply-demand ratio R: where R eq is the supply-demand balance threshold, and γ is the adjustment coefficient.

8. The real-time data-driven demand response bus intelligent scheduling method according to claim 1, characterized in that, In S6, based on the passenger's historical travel data and real-time preferences, the collaborative filtering algorithm is used to recommend the optimal route, and a personalized ride plan is pushed through the APP to improve satisfaction.

9. A real-time data-driven demand response bus intelligent scheduling system, characterized in that, It includes the following modules: Multi-source data collection and fusion module, used to collect and fuse multi-source heterogeneous data in real time; Demand prediction and clustering module, implementing short-term demand prediction and clustering based on LSTM and DBSCAN algorithms; Dynamic planning and scheduling module, using the Voronoi diagram and greedy algorithm to generate dynamic routes; Vehicle scheduling optimization module, optimizing the route through mixed integer programming and simulated annealing algorithm; Fare dynamic management module, adjusting the real-time fare based on the supply-demand game model; User service matching module, pushing personalized solutions through the collaborative filtering algorithm; Simulation verification module, verifying the scheduling plan on the digital twin platform; Feedback optimization module, implementing closed-loop parameter optimization based on reinforcement learning.

10. The real-time data-driven demand response bus intelligent scheduling system according to claim 9, characterized in that The feedback optimization module receives the inputs from the simulation verification module and the user service matching module, dynamically adjusts the scheduling parameters, and feeds them back to the dynamic planning and scheduling module.

Images