A method and system for collaborative scheduling of online car-hailing and public transportation based on a cloud scheduling platform

Abstract

This application provides a method and system for coordinated scheduling of ride-hailing and public transportation based on a cloud scheduling platform, belonging to the field of coordinated scheduling technology. The method involves: determining a connection response time window; further filtering candidate connection stops and determining their connection priorities; generating connection routes; determining the collaborative gain between candidate connection stops and their connection priorities based on the route's cost and the priority of the candidate connection stops; assessing the off-peak risk of transfer timing by associating the arrival reliability of bus services with the arrival margin of ride-hailing connection routes, thus obtaining a transfer off-peak risk factor; and jointly rolling optimization of candidate connection stops and connection routes under negative risk constraints to obtain target connection stops and target connection routes. This application enables coordinated optimization of connection points and routes in dynamic connection scenarios, thereby improving the efficiency of coordinated connections.

Description

Technical Field

[0001] This application relates to the field of collaborative scheduling technology, and more specifically, to a method and system for collaborative scheduling of ride-hailing and public transportation based on a cloud scheduling platform. Background Technology

[0002] With the continuous development of urban public transportation systems and shared mobility services, collaborative scheduling technology has gradually become an important technical means to improve the comprehensive mobility service capabilities. Collaborative scheduling technology can coordinate and manage various transportation resources in a unified manner. By integrating vehicle operating status, passenger travel needs, and traffic network information, it can achieve efficient connection and collaborative operation between different modes of transportation. In scenarios such as ride-hailing connecting to buses, rail transit connecting to buses, and transfers at integrated transportation hubs, collaborative scheduling technology can improve the efficiency of transportation resource utilization, optimize passenger travel experience, and provide important technical support for building a convenient and efficient integrated urban mobility system.

[0003] In existing technologies, ride-hailing and public transport coordination typically requires considering bus arrival times, passenger distribution, and road conditions to make scheduling decisions. However, due to the continuous changes in road network congestion, passenger order locations, vehicle trajectories, and bus arrival times, the coordination process is highly dynamic and uncertain. Most existing scheduling systems either pre-set connection points or handle connection point selection and vehicle route planning separately, lacking a comprehensive consideration of the relationship between the suitability of connection points and vehicle traffic conditions. When road conditions change, existing connection points may no longer provide adequate connectivity; when connection points are adjusted, existing vehicle routes may not meet public transport transfer needs, easily leading to increased vehicle detours, longer passenger waiting times, and mismatched connection sequences, reducing the overall operational efficiency of ride-hailing and public transport coordination. Therefore, how to achieve collaborative optimization of connection points and routes in dynamic coordination scenarios to improve coordination efficiency has become a challenge for the industry. Summary of the Invention

[0004] This application provides a method and system for coordinated scheduling of ride-hailing and public transportation based on a cloud scheduling platform, which can realize the coordinated optimization of connection points and travel routes in dynamic connection scenarios, thereby improving the efficiency of coordinated connection.

[0005] Firstly, this application provides a method for coordinated dispatching of ride-hailing and public transportation based on a cloud dispatching platform, comprising the following steps: The cloud dispatch platform obtains the road network traffic status of the target connection area and the estimated arrival time of the associated bus routes, and determines the connection response time window through the estimated arrival time; Candidate shuttle stops are selected based on the order location of passengers to be picked up and the road network traffic status, and the pick-up priority of the candidate shuttle stops is determined by the spatiotemporal convergence of passenger flow within the pick-up response time window. A connecting route is generated based on the real-time location of the ride-hailing vehicle and candidate pick-up stops. The collaborative gain between the candidate pick-up stops and the connecting route is determined by the toll cost of the connecting route and the pick-up priority of the candidate pick-up stops. By associating the arrival reliability of bus routes with the arrival margin of ride-hailing connection routes, the risk of staggered transfer times is assessed, and the staggered transfer risk factor is obtained. By combining the synergistic gain and the transfer off-peak risk factor, the candidate connecting stops and connecting routes are jointly optimized under negative risk constraints to obtain the target connecting stops and target connecting routes. The target shuttle stop and target shuttle route are sent to the ride-hailing terminal and the bus dispatch terminal.

[0006] In this embodiment, determining the connection response time window through the estimated arrival time specifically includes: Obtain the estimated arrival time of the associated bus routes at the corresponding bus stops within the target connection area; Based on the estimated arrival time, extract the base arrival time of the bus service entering the target connection area; Based on the aforementioned arrival reference time, a buffer zone for advance pick-up of ride-hailing vehicles and a permissible waiting period for passenger transfers are set. A connection response time window is generated based on the advance connection buffer segment and the transfer waiting allowance segment.

[0007] In this embodiment, the process of filtering candidate shuttle stops based on the order location of the passengers to be picked up and the road network traffic status specifically includes: Extract spatial clusters of orders within the target pick-up area based on the order locations of passengers to be picked up; The initial set of shuttle stops is determined based on the matching relationship between the order space clusters and the associated bus stops in terms of their connection radius. Based on the road network traffic status, the initial set of connecting stops is checked for road segment traffic constraints to obtain a set of reachable stops that meet the connecting conditions. Candidate shuttle stops are selected based on the temporary parking capacity conditions of each stop in the set of reachable stops and the passenger pedestrian access boundary.

[0008] In this embodiment, determining the connection priority of candidate shuttle stops based on the spatiotemporal convergence of passenger flow within the connection response time window specifically includes: Extract the arrival sequence of pending shuttle orders corresponding to the candidate shuttle stops based on the shuttle response time window; Based on the arrival sequence of the pending shuttle orders, the concentrated arrival segments of the orders are identified, thereby determining the passenger flow convergence period characteristics of the candidate shuttle stops; Based on the spatial distribution relationship between the order locations of passengers to be transferred and the candidate transfer stops, the spatial aggregation features of passenger flow at the candidate transfer stops are extracted; Based on the passenger flow convergence time period characteristics and the passenger flow spatial convergence characteristics, the spatiotemporal convergence degree of passenger flow at the candidate shuttle stop within the shuttle response time window is determined; The candidate shuttle stops are prioritized by the spatiotemporal convergence of passenger flow to obtain the shuttle priority of the candidate shuttle stops.

[0009] In this embodiment, generating a shuttle route based on the real-time location of the ride-hailing vehicle and candidate pick-up points specifically includes: Obtain the real-time location of ride-hailing vehicles within the target pick-up area; Based on the real-time location and each candidate pick-up stop, establish the pick-up route pointing relationship from the ride-hailing vehicle to the candidate pick-up stop; Based on the connection path pointing relationship and the road network traffic status, a sequence of passable road segments that meet the connection response time window requirements is selected. Based on the sequence of passable road segments, a connecting route is generated for ride-hailing vehicles to reach candidate pick-up points.

[0010] In this embodiment, determining the cooperative gain between a candidate shuttle stop and the shuttle route based on the toll cost of the shuttle route and the shuttle priority of the candidate shuttle stop specifically includes: Determine the toll cost of connecting routes; Extract the connection revenue characteristics of candidate shuttle stops based on their connection priority; Based on the passage cost and the connection benefit characteristics, a collaborative gain analysis is performed on the candidate connection stops and connection routes to obtain the collaborative gain degree between the candidate connection stops and connection routes.

[0011] In this embodiment, the risk assessment of staggered transfer connection timing is conducted by associating the arrival reliability of bus routes with the arrival margin of ride-hailing connection routes. The specific staggered transfer risk factors include: Determine the arrival reliability of associated bus routes; The arrival time confidence region of ride-hailing vehicles is determined based on the arrival margin of the ride-hailing connection route; Based on the arrival reliability of associated bus routes and the arrival time confidence region of ride-hailing vehicles, a time connection conflict analysis is performed under the condition of link breakage triggering to obtain the conditional risk field of transfer connection link breakage. The passenger flow at candidate connecting stops is assessed for off-peak risk using the conditional risk field to obtain the off-peak transfer risk factor.

[0012] In this embodiment, the joint rolling optimization of candidate connecting stops and connecting routes by applying negative risk constraints through the synergistic gain degree and the transfer off-peak risk factor specifically includes: Using candidate shuttle stops and shuttle routes as decision variables, the synergistic gain degree as a positive optimization driver, and the transfer off-peak risk factor as a negative risk constraint, a joint optimization space is constructed. Negative risk filtering is performed on the joint optimization space to eliminate candidate solutions whose transfer peak-shifting risk factors exceed the dynamic risk threshold, thereby obtaining a feasible solution domain that satisfies the negative risk constraint. Within the feasible solution domain, rolling optimization is performed with the cooperative gain degree as the main objective, the gain ranking of candidate solutions is updated, and the negative risk boundary is simultaneously verified. The candidate solution sequence of rolling optimization is balanced between gain and negative risk to obtain the target docking point and the target shuttle route.

[0013] In this embodiment, the rolling optimization within the feasible solution domain with the cooperative gain as the primary objective, updating the gain ranking of candidate solutions, and synchronously verifying the negative risk boundary specifically includes: Within the feasible solution domain, the cooperative gain degree is converted into the gain potential energy of the candidate solution, and the candidate solution is selected by rolling selection along the gain potential energy gradient direction to obtain the candidate solution sequence in the current time domain. For each solution in the candidate solution sequence, a negative risk boundary penetration determination is performed, and the candidate solutions are weighted by gain reduction based on the penetration depth to generate a weighted gain set; The gain weighted set is reintegrated into the rolling selection process within the feasible solution domain. The gain ranking of the previous rolling time domain is used as the warm start condition to generate an updated gain ranking and proceed to the next rolling time domain.

[0014] Secondly, this application provides a ride-hailing and public transport collaborative dispatching system based on a cloud dispatching platform, used to execute a ride-hailing and public transport collaborative dispatching method based on a cloud dispatching platform. The collaborative dispatching system includes: The connection response time window generation module is used by the cloud scheduling platform to obtain the road network traffic status of the target connection area and the estimated arrival time of the associated bus routes, and to determine the connection response time window through the estimated arrival time; The candidate shuttle stop evaluation module is used to filter candidate shuttle stops based on the order location of passengers to be picked up and the road network traffic status, and to determine the shuttle priority of candidate shuttle stops by the spatiotemporal convergence of passenger flow within the shuttle response time window; The connecting path coordination gain module is used to generate connecting routes based on the real-time location of ride-hailing vehicles and candidate connecting stops, and to determine the coordination gain between candidate connecting stops and connecting routes by the toll cost of the connecting routes and the connecting priority of candidate connecting stops. The transfer off-peak risk assessment module is used to assess the off-peak risk of transfer connection sequence by associating the arrival reliability of bus routes and the arrival margin of ride-hailing connection routes, and obtain the transfer off-peak risk factor. The joint rolling optimization decision module is used to jointly roll optimize the candidate connecting stops and connecting routes by applying negative risk constraints to the collaborative gain degree and the transfer off-peak risk factor, so as to obtain the target connecting stops and the target connecting routes. The collaborative scheduling scheme distribution module is used to distribute the target shuttle stop points and target shuttle routes to ride-hailing terminals and bus dispatch terminals.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: The cloud dispatch platform obtains the road network traffic status of the target connection area and the estimated arrival time of associated bus routes, and determines the connection response time window based on the estimated arrival time; it filters candidate connection stops based on the order location of passengers to be connected and the road network traffic status, and determines the connection priority of candidate connection stops based on the spatiotemporal convergence of passenger flow within the connection response time window; it generates a connection route based on the real-time location of ride-hailing vehicles and candidate connection stops, and determines the connection priority of candidate connection stops based on the toll cost of the connection route and the connection speed of the candidate connection stops. The priority of the transfer determines the synergistic gain between candidate transfer stops and transfer routes; the arrival reliability of associated bus routes and the arrival margin of ride-hailing transfer routes are used to assess the off-peak risk of transfer connection timing, resulting in a transfer off-peak risk factor; the candidate transfer stops and transfer routes are jointly and continuously optimized under negative risk constraints using the synergistic gain and the transfer off-peak risk factor to obtain the target transfer stops and target transfer routes; the target transfer stops and target transfer routes are then distributed to ride-hailing terminals and bus dispatch terminals.

[0016] Therefore, this application demonstrates that the target shuttle stop and target shuttle route can be distributed to ride-hailing terminals and bus dispatch terminals. Firstly, by acquiring the road network traffic status of the target shuttle area and the estimated arrival times of associated bus routes, and determining the shuttle response time window accordingly, subsequent dispatching can be established within the time frame jointly defined by road traffic changes and bus arrival processes, preventing shuttle arrangements from deviating from the actual transfer sequence. Secondly, by determining the shuttle priority based on the order location of passengers awaiting shuttle access, road network traffic status, and the spatiotemporal convergence of passenger flow at candidate shuttle stops within the shuttle response time window, shuttle priority can be determined, balancing passenger convergence needs and road accessibility, filtering out stops with insufficient passenger capacity or poor traffic conditions, and reducing passenger dispersion, vehicle detours, and extended waiting times caused by fixed-point dispatching. Furthermore, the shuttle priority is determined by the toll cost of the shuttle route and the shuttle priority of candidate shuttle stops. The collaborative gain factor incorporates the passenger carrying value of a stop and the transit cost of a ride-hailing vehicle to that stop into the same evaluation process, avoiding the separation between stop selection and vehicle route planning. It also assesses the risk of peak-hour transfers by combining the arrival reliability of bus services and the arrival margin of ride-hailing routes, identifying candidate combinations with low transit costs but mismatched transfer timings. Finally, through the collaborative gain factor and peak-hour transfer risk factors, a joint rolling optimization of candidate stop locations and routes is performed with negative risk constraints. This allows for synchronized adjustment of stop locations and routes based on changes in road network traffic conditions, ride-hailing vehicle locations, and bus arrival status. The scheduling results are then distributed to ride-hailing terminals and bus dispatch terminals, reducing route mismatches after stop adjustments and transfer timing mismatches after route changes. This, in turn, reduces vehicle detour distances and passenger waiting times, improving the overall operational efficiency of the ride-hailing and bus connection process.

[0017] In summary, the technical solution adopted in this application can achieve collaborative optimization of connection points and travel routes in dynamic connection scenarios, thereby improving collaborative connection efficiency. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this embodiment of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is an exemplary flowchart of a cloud-based dispatching platform-based collaborative dispatching method for ride-hailing and public transportation, provided in this application. Figure 2 This is a schematic diagram based on the verification results of the spatiotemporal convergence of passenger flow provided in this application; Figure 3 This is based on the distribution map of transfer off-peak risk factors provided in this application; Figure 4 This is a modular structure diagram of a ride-hailing and public transport collaborative dispatch system based on a cloud dispatch platform, provided in this application. Figure 5 This is a schematic diagram illustrating the application scenario of ride-hailing and public transportation collaborative connection under the cloud dispatch platform provided in this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] This application provides a method and system for coordinated dispatching of ride-hailing and public transportation based on a cloud dispatch platform. The core of this system is that the cloud dispatch platform obtains the road network traffic status of the target connection area and the estimated arrival times of associated bus routes, and determines the connection response time window based on the estimated arrival times. Candidate connection stops are selected based on the order location of the passengers to be connected and the road network traffic status, and the connection priority of the candidate connection stops is determined by the spatiotemporal convergence of passenger flow within the connection response time window. A connection route is generated based on the real-time location of the ride-hailing vehicle and the candidate connection stops, and the connection is then established. The synergistic gain between candidate connecting stops and connecting routes is determined by the travel cost of the route and the connection priority of candidate connecting stops. Peak-hour risk assessment of transfer connection timing is conducted by associating the arrival reliability of bus services with the arrival margin of ride-hailing connecting routes, resulting in a peak-hour transfer risk factor. The candidate connecting stops and connecting routes are then jointly and continuously optimized under negative risk constraints using the synergistic gain and the peak-hour transfer risk factor to obtain target connecting stops and target connecting routes. The target connecting stops and target connecting routes are then distributed to ride-hailing terminals and bus dispatch terminals.

[0022] Example 1 To better understand the above technical solutions, a detailed description will be provided below with reference to the accompanying drawings and specific implementation methods. Figure 1 As shown in the figure, this is an exemplary flowchart of a ride-hailing and public transport collaborative scheduling method based on a cloud scheduling platform according to this embodiment of the application, including the following steps: In step S1, the cloud dispatch platform obtains the road network traffic status of the target connection area and the estimated arrival time of the associated bus routes, and determines the connection response time window through the estimated arrival time.

[0023] It should be noted that the cloud dispatch platform mentioned in this application refers to a cloud processing platform used for the coordinated dispatch of ride-hailing vehicles and public buses; the road network traffic status refers to status information used to reflect the current traffic conditions of each road segment within the target connection area; and the estimated arrival time refers to the estimated arrival time of the associated bus service at the corresponding bus stop within the target connection area.

[0024] In this embodiment, determining the connection response time window based on the estimated arrival time can be achieved through the following steps: Obtain the estimated arrival time of the associated bus routes at the corresponding bus stops within the target connection area; Based on the estimated arrival time, extract the base arrival time of the bus service entering the target connection area; Based on the aforementioned arrival reference time, a buffer zone for advance pick-up of ride-hailing vehicles and a permissible waiting period for passenger transfers are set. A connection response time window is generated based on the advance connection buffer segment and the transfer waiting allowance segment.

[0025] It should be noted that the arrival reference time mentioned in this application refers to the time when the bus enters the target connection area as the reference for connection time arrangement; the advance connection buffer period refers to the time period reserved for ride-hailing vehicles to arrive at the connection stop in advance before the bus arrives; the transfer waiting allowance period refers to the time period during which passengers are allowed to wait to complete the transfer at the connection stop; and the connection response time window refers to the allowable time range used to constrain ride-hailing vehicle connection and passenger transfer.

[0026] In practice, firstly, the estimated arrival times of the associated bus routes within the target transfer area are retrieved. If the bus route passes through multiple bus stops within the target transfer area, the estimated arrival times of each stop are sorted in order, and the earliest estimated arrival time is taken as the baseline arrival time for the bus route entering the target transfer area. Secondly, using the baseline arrival time as a reference point, a preset buffer time is decremented backwards. This preset buffer time can be set based on the average travel time of the ride-hailing vehicle within the area and the preparation time for temporary stops at designated stops. The time period consisting of the arrival reference time and the departure time is used as the advance pick-up buffer period for ride-hailing services. At the same time, a preset allowable time is extended from the arrival reference time. The preset allowable time can be set according to the average walking time and waiting tolerance time of passengers after getting off the vehicle to the pick-up stop. The time period consisting of the arrival reference time and the extended time is used as the passenger's transfer waiting allowable period. Finally, the start time of the advance pick-up buffer period is used as the lower limit of the time window, the end time of the transfer waiting allowable period is used as the upper limit of the time window, and the continuous time range covered by the lower limit to the upper limit is used as the pick-up response time window.

[0027] In step S2, candidate shuttle stops are selected based on the order location of the passengers to be connected and the road network traffic status, and the connection priority of the candidate shuttle stops is determined by the spatiotemporal convergence of passenger flow within the connection response time window.

[0028] In this embodiment, the selection of candidate shuttle stops based on the order location of the passengers to be picked up and the road network traffic status can be achieved through the following steps: Extract spatial clusters of orders within the target pick-up area based on the order locations of passengers to be picked up; The initial set of shuttle stops is determined based on the matching relationship between the order space clusters and the associated bus stops in terms of their connection radius. Based on the road network traffic status, the initial set of connecting stops is checked for road segment traffic constraints to obtain a set of reachable stops that meet the connecting conditions. Candidate shuttle stops are selected based on the temporary parking capacity conditions of each stop in the set of reachable stops and the passenger pedestrian access boundary.

[0029] It should be noted that, in this application, the order spatial cluster refers to the order distribution cluster formed by the spatial convergence of passenger orders to be connected that are geographically close to each other; the initial screening set of connecting stops refers to the set of candidate stops that meet the connection radius matching relationship with the order spatial cluster; the set of reachable stops refers to the set of stops in the initial screening set of connecting stops that meet the connection reachability conditions; and the candidate connecting stops refers to connecting stops that simultaneously meet the temporary stop carrying conditions and the passenger pedestrian access boundary.

[0030] In specific implementation, firstly, the density-based DBSCAN clustering algorithm is used to cluster the order locations of passengers to be connected. The neighborhood radius parameter is set as the acceptable walking distance for passengers to connect, where this spatial range can be set according to the acceptable walking distance from passengers to the stop. The minimum number of orders parameter is set as the lower limit of orders constituting effective connection demand. The order clusters formed by the aggregation of neighboring orders output by the clustering algorithm are taken as the order spatial clusters within the target connection area. Secondly, the connection coverage area is delineated with the geometric center of each order spatial cluster as the center and a preset connection radius (which can be set according to the reasonable passenger pick-up distance for ride-hailing short-distance connections) as the radius. The spatial distance from each associated bus stop to the geometric center of the order spatial cluster is calculated one by one. The candidate stops corresponding to bus stops with a spatial distance not greater than the connection radius are collected one by one. The set of these candidate stops that meet the connection radius matching relationship is taken as the initial screening set of connection stops. Then, for the connection... For each stop in the initial screening set of pick-up stops, based on the average driving speed and road occupancy rate of each road segment in the road network traffic status, each stop is checked to see if there are any prohibited or severely congested road segments along the route from the ride-hailing vehicle's real-time location to that stop where the vehicle's speed is below the traffic threshold or the road occupancy rate is above the congestion threshold. Stops without such prohibited or severely congested road segments and with the conditions for pick-up access are retained one by one, and the set of these retained stops is called the set of accessible stops. Finally, for each stop in the set of accessible stops, each stop is checked to see if the number of temporary parking spaces at that stop is not less than the number of ride-hailing vehicles to be picked up, and if the walking distance from that stop to the geometric center of the order space cluster does not exceed the passenger walking access boundary. The passenger walking access boundary can be set according to the maximum walking access distance acceptable to passengers. Stops that simultaneously meet the conditions of the number of temporary parking spaces and the walking access boundary are screened out one by one, and these screened stops are called candidate pick-up stops.

[0031] In this embodiment, determining the connection priority of candidate shuttle stops based on the spatiotemporal convergence of passenger flow within the connection response time window can be achieved through the following steps: Extract the arrival sequence of pending shuttle orders corresponding to the candidate shuttle stops based on the shuttle response time window; Based on the arrival sequence of the pending shuttle orders, the concentrated arrival segments of the orders are identified, thereby determining the passenger flow convergence period characteristics of the candidate shuttle stops; Based on the spatial distribution relationship between the order locations of passengers to be transferred and the candidate transfer stops, the spatial aggregation features of passenger flow at the candidate transfer stops are extracted; Based on the passenger flow convergence time period characteristics and the passenger flow spatial convergence characteristics, the spatiotemporal convergence degree of passenger flow at the candidate shuttle stop within the shuttle response time window is determined; The candidate shuttle stops are prioritized by the spatiotemporal convergence of passenger flow to obtain the shuttle priority of the candidate shuttle stops.

[0032] It should be noted that, in this application, the order arrival sequence to be connected refers to the sequence of orders to be connected from candidate connecting stops arranged chronologically within the connecting response time window; the passenger flow convergence period characteristic refers to the period when passenger flow arrives at candidate connecting stops in a concentrated manner within the connecting response time window; the passenger flow spatial aggregation characteristic refers to the spatial clustering and distribution of passenger orders to be connected around candidate connecting stops; the passenger flow spatiotemporal convergence degree refers to the degree of convergence of passenger flow at candidate connecting stops in time and space within the connecting response time window; and the connecting priority indicates the level at which candidate connecting stops are preferentially selected for connecting.

[0033] In practice, firstly, using the lower and upper bounds of the shuttle response time window as time boundaries, all pending shuttle orders falling within this time range and whose passenger walking access boundary covers the candidate shuttle stop are retrieved from the order database. These orders are then sorted from earliest to latest by their order placement time, and the sorted order sequence is used as the arrival sequence of pending shuttle orders corresponding to that candidate shuttle stop. Secondly, a sliding time window counting method is used for the arrival sequence of pending shuttle orders. The width of the statistical window is set to a time granularity that reflects passenger flow concentration. The time granularity can be set according to the duration of the shuttle response time window and the accuracy of passenger flow statistics. The number of orders falling within each statistical window is counted window by window, and orders exceeding this number are counted. The statistical window that exceeds the threshold for concentrated arrival is marked as the concentrated arrival segment of orders. The characteristics of the time periods covered by each concentrated arrival segment of orders are used as the passenger flow aggregation time characteristics of the candidate shuttle stop. Next, the spatial distance from the order location of each passenger to be transferred to the candidate shuttle stop is calculated one by one. An inverse distance weighting method is used to assign a spatial weight to each order location, with the closer the distance, the greater the weight. The sum of the spatial weights of each order location is used as the value reflecting the density of order spatial aggregation around the stop as the passenger flow spatial aggregation characteristic of the candidate shuttle stop. Then, the spatiotemporal aggregation degree of passenger flow at the candidate shuttle stop within the shuttle response time window is calculated using the following formula: in, Indicates the first The spatiotemporal convergence of passenger flow at each candidate shuttle stop within the shuttle response time window; Indicates the first Characteristics of passenger flow convergence periods at candidate shuttle stops; This represents the minimum value among the passenger flow convergence time characteristics of each candidate shuttle stop; This represents the maximum value among the passenger flow convergence time characteristics of each candidate shuttle stop; Indicates the first Passenger flow spatial aggregation characteristics of candidate shuttle stops; This represents the minimum value among the spatial aggregation characteristics of passenger flow at each candidate shuttle stop; This represents the maximum value among the spatial aggregation characteristics of passenger flow for each candidate shuttle stop; This represents the time dimension weighting coefficient corresponding to the characteristics of the peak passenger flow period; This represents the spatial dimension weight coefficient corresponding to the spatial aggregation feature of passenger flow, and ; This indicates an extremely small positive number used to avoid a denominator of zero; finally, the candidate connecting stops are sorted from largest to smallest according to the spatiotemporal convergence of passenger flow, and the sorting position is used as the connection priority of each candidate connecting stop. The larger the spatiotemporal convergence of passenger flow, the higher the connection priority.

[0034] For example, order arrival records, passenger walking access distance, and stop conditions within the same connection response time window in the target connection area can be selected as verification samples. Table 1 provides exemplary statistical results for four candidate connection stops. The proportion of concentrated arrival orders is used to reflect the characteristics of passenger flow convergence time periods, the average walking access distance is used to reflect the convenience of passenger access, the spatial aggregation component is used to reflect the degree of aggregation of order locations around the stop, and the spatiotemporal convergence degree of passenger flow is used to participate in the connection priority ranking.

[0035] Table 1. Verification data on the spatiotemporal convergence of passenger flow at candidate shuttle stops. As shown in Table 1, stop A had 21 orders within the connection response time window, with a concentrated arrival order ratio of 0.71, an average walking access distance of 176m, a spatial aggregation component of 0.88, and a passenger flow spatiotemporal convergence of 0.81, ranking first in priority. Although stop D met the connection access conditions, its number of orders, concentrated arrival order ratio, and spatial aggregation component were all lower than other stops, with a passenger flow spatiotemporal convergence of 0.44, ranking fourth in priority. The above data shows that relying solely on the number of orders or spatial distance is insufficient to fully reflect the strength of connection demand. However, by incorporating both the time-concentrated arrival characteristics and spatial aggregation characteristics into the passenger flow spatiotemporal convergence, the priority of candidate connection stops can be kept consistent with the actual distribution of connection demand.

[0036] refer to Figure 2 As shown, Figure 2 This is a schematic diagram illustrating the verification results of the spatiotemporal convergence of passenger flow provided in this application. Figure 2Among them, stop A maintained the highest level in terms of time convergence component, spatial convergence component, and passenger flow spatiotemporal convergence degree, indicating that stop A has good connection organization conditions in both time concentration and spatial access convenience. Stops B and C have similar but still different passenger flow spatiotemporal convergence degrees, which can provide a quantitative basis for the ranking of stops within the same connection response time window. Stop D has all three indicators at a low level, making it suitable as an alternative connection stop rather than a priority target.

[0037] In step S3, a connecting route is generated based on the real-time location of the ride-hailing vehicle and the candidate connecting stops. The cooperative gain between the candidate connecting stops and the connecting route is determined by the travel cost of the connecting route and the connecting priority of the candidate connecting stops.

[0038] In this embodiment, generating a shuttle route based on the real-time location of the ride-hailing vehicle and candidate pick-up points can be achieved through the following steps: Obtain the real-time location of ride-hailing vehicles within the target pick-up area; Based on the real-time location and each candidate pick-up stop, establish the pick-up route pointing relationship from the ride-hailing vehicle to the candidate pick-up stop; Based on the connection path pointing relationship and the road network traffic status, a sequence of passable road segments that meet the connection response time window requirements is selected. Based on the sequence of passable road segments, a connecting route is generated for ride-hailing vehicles to reach candidate pick-up points.

[0039] It should be noted that, in this application, the real-time location refers to the current location of the ride-hailing vehicle within the target pick-up area; the pick-up path pointing relationship refers to the path connection relationship of the ride-hailing vehicle from the real-time location to the candidate pick-up stop; the passable road segment sequence refers to the sequence of road segments that meet the pick-up response time window requirements and are passable by the ride-hailing vehicle; and the pick-up travel path refers to the travel path of the ride-hailing vehicle from the real-time location to the candidate pick-up stop.

[0040] In practice, firstly, the GPS positioning coordinates uploaded by the ride-hailing vehicle's onboard terminal according to a preset reporting cycle are retrieved. The location of the road segment where these GPS positioning coordinates fall after map matching in the target pick-up area road network is taken as the real-time location of the ride-hailing vehicle in the target pick-up area. Secondly, using the road segment node where the ride-hailing vehicle's real-time location is located as the starting node and the road segment nodes where each candidate pick-up stop is located as the ending node, directed connections from the starting node to each ending node are established one by one in the road network topology map of the target pick-up area. These directed connections from the ride-hailing vehicle's real-time location to each candidate pick-up stop are taken as the pick-up path pointing relationship from the ride-hailing vehicle to the candidate pick-up stop. Then, the travel time of each road segment is calculated based on the average driving speed of each road segment in the road network traffic status, and the travel time is calculated along the pick-up path. The pointing relationship uses Dijkstra's shortest path algorithm, with the travel time of each road segment as the weight, to search for connected road segments from the starting node to each ending node. It verifies whether the cumulative travel time of each searched connected road segment allows the ride-hailing vehicle to reach the candidate pick-up stop within the advance pick-up buffer period of the pick-up response time window. Connected road segments whose cumulative travel time meets the time limit requirement of the advance pick-up buffer period are arranged in order of passage, and the arranged road segment sequence is taken as the sequence of passable road segments that meet the requirements of the pick-up response time window. Finally, the adjacent road segments in the passable road segment sequence are connected end to end in the road network topology map in order of passage, and the continuous passage route extending from the real-time location of the ride-hailing vehicle to the candidate pick-up stop after the connection is taken as the pick-up passage path of the ride-hailing vehicle to the candidate pick-up stop.

[0041] In this embodiment, determining the cooperative gain between a candidate shuttle stop and a shuttle route by using the travel cost of the shuttle route and the shuttle priority of the candidate shuttle stop can be achieved through the following steps: Determine the toll cost of connecting routes; Extract the connection revenue characteristics of candidate shuttle stops based on their connection priority; Based on the passage cost and the connection benefit characteristics, a collaborative gain analysis is performed on the candidate connection stops and connection routes to obtain the collaborative gain degree between the candidate connection stops and connection routes.

[0042] It should be noted that the passage cost mentioned in this application refers to the cost incurred by a ride-hailing vehicle to travel along the connecting passage route; the connecting revenue characteristic refers to the revenue characteristic that can be brought about by the selection of candidate connecting stops for connecting; and the synergistic gain degree refers to the comprehensive gain generated by the cooperation between candidate connecting stops and connecting passage routes.

[0043] In specific implementation, firstly, the total mileage of the connecting route is obtained by accumulating the lengths of each road segment in the order of passage. Then, the total travel time of each road segment is accumulated based on the average speed of each segment in the road network traffic conditions. The total mileage and total travel time are then weighted and summed according to a preset cost weight, and the resulting weighted sum is used as the toll cost of the connecting route. Secondly, based on the connecting priority of candidate connecting stops, higher priority points correspond to higher revenue values, which are mapped and assigned. The connecting priority ranks, from high to low, correspond to revenue values ​​in the range of 1.0 to 0.1. The values ​​reflecting the connecting revenue of the stops are used as the connecting revenue characteristics of the candidate connecting stops. Finally, based on the connecting revenue characteristics of the candidate connecting stops, the collaborative gain between the candidate connecting stops and the connecting route is calculated using the following formula: in, Indicates the first The candidate shuttle stop and the first The degree of collaborative gain between connecting routes; Indicates the first The connection revenue characteristics of each candidate connection stop; and These represent the minimum and maximum connection revenue characteristics in the candidate connection stop set, respectively; Indicates the first The connecting route reaches the first Total route mileage to each candidate shuttle stop; Indicates the first The connecting route reaches the first Total path time for each candidate shuttle stop; and These represent the minimum and maximum total path mileage in each connecting route, respectively. and These represent the minimum total time and the maximum total time for each connecting route, respectively. This indicates the contribution weight of connection revenue characteristics to the synergistic gain. It can be set based on the density of pending connection orders and the intensity of bus arrival transfer demand within the target connection area. When the order density is high or the transfer demand is concentrated, the contribution weight is increased. The value of is selected to prioritize candidate docking points with higher connection benefits; This represents the suppression weight of toll cost on the degree of cooperative gain. It can be set according to the degree of road network congestion in the target connection area and the urgency of the connection response time window. When the degree of road network congestion is high or the connection response time window is short, the weight is increased. The value of is chosen to reduce the probability that high-cost paths are selected; This represents the weighting coefficient of the total route mileage in the toll cost. It can be set according to the relative importance of the total route mileage and total route time in influencing shuttle service. When vehicle travel costs or detour distance control requirements are high, this coefficient should be increased. The value of should be reduced when the on-time requirement is high. The value of the path and the corresponding increase in the impact of the total path time component; This represents a very small positive number used to avoid a denominator of zero.

[0044] In step S4, the risk assessment of peak-shifting for transfer connection timing is conducted by associating the arrival reliability of bus routes with the arrival margin of ride-hailing connection routes, thus obtaining the peak-shifting risk factor for transfer.

[0045] In this embodiment, the risk assessment of off-peak transfer connection timing by associating the arrival reliability of bus routes with the arrival margin of ride-hailing connection routes can be achieved through the following steps: Determine the arrival reliability of associated bus routes; The arrival time confidence region of ride-hailing vehicles is determined based on the arrival margin of the ride-hailing connection route; Based on the arrival reliability of associated bus routes and the arrival time confidence region of ride-hailing vehicles, a time connection conflict analysis is performed under the condition of link breakage triggering to obtain the conditional risk field of transfer connection link breakage. The passenger flow at candidate connecting stops is assessed for off-peak risk using the conditional risk field to obtain the off-peak transfer risk factor.

[0046] It should be noted that, in this application, the arrival reliability refers to the degree of confidence that the associated bus will arrive at the corresponding station according to the expected arrival time; the arrival margin refers to the time buffer margin used by the ride-hailing vehicle to offset the fluctuation of the road network traffic status and ensure timely arrival at the candidate pick-up stop when traveling along the connecting route; the arrival time confidence region refers to the confidence interval of the time fluctuation of the ride-hailing vehicle arriving at the candidate pick-up stop; the conditional risk field of the transfer connection failure refers to the continuous distribution field of the risk intensity of the time connection conflict between the bus and the ride-hailing vehicle at each spatiotemporal node under the condition of the failure trigger; and the transfer off-peak risk factor refers to the quantitative value of the risk of the transfer failure caused by the mismatch of the time sequence of the bus and the ride-hailing vehicle at the candidate pick-up stop.

[0047] In practice, the deviation between the arrival time and the expected arrival time of related bus routes at the same time within the past 30 days is statistically analyzed. The percentage of bus routes with a deviation within ±2 minutes is used as the arrival reliability of the related bus routes. The connecting route is broken down into continuous road segment units, and the standard deviation of the travel time for each segment at the same time within the past 7 days is statistically analyzed. The time corresponding to three times the standard deviation is taken as the single-segment travel margin. The three-times standard deviation is set according to the 99.7% confidence level of the normal distribution and can cover most traffic fluctuation scenarios. The single-segment travel margins of all road segments on the route are added together, and the total time obtained is used as the arrival margin of the ride-hailing connecting route. The expected travel time of the ride-hailing connecting route is used as the mean, and the arrival margin is taken as the time margin corresponding to three times the standard deviation. The time range corresponding to one standard deviation is extended forward and backward along the expected arrival time, and this time range is taken as the arrival time confidence region of the ride-hailing. A spatiotemporal grid partitioning method is used. A two-dimensional spatiotemporal grid is constructed using time nodes divided into 15-second intervals within the connection response time window as the horizontal axis and the spatial location sequence of each candidate connection stop as the vertical axis. For each grid node, the bus arrival fluctuation interval is obtained by extracting the upper and lower limits of the historical arrival deviation time based on the quantile corresponding to the arrival reliability, with the bus arrival fluctuation interval as the center and the arrival time confidence region of the corresponding ride-hailing route. The proportion of the non-overlapping time of the two intervals to the total time of the connection response time window is calculated, and this proportion is used as the chain break risk intensity of the spatiotemporal node. The chain break risk intensities of all spatiotemporal nodes together constitute the conditional risk field of transfer connection chain breakage. The spatiotemporal convergence degree of passenger flow of the candidate connection stops is matched to each time node of the corresponding stop in the conditional risk field according to the time node. The chain break risk intensity of each time node is multiplied by the corresponding passenger flow proportion and then summed. The weighted sum obtained is used as the transfer off-peak risk factor.

[0048] For example, by replaying bus arrival records, ride-hailing route passage records, and candidate stop passenger flow records within the same target connection area, the reliability of bus arrivals was set to 0.72, 0.80, 0.88, and 0.94, and the arrival margin of ride-hailing connection routes was set to 1 minute, 2 minutes, 3 minutes, 4 minutes, and 5 minutes. The transfer off-peak risk factors under different combinations were obtained according to the conditional risk field calculation method of this application. (Reference) Figure 3As shown in the figure, this is a distribution map of the risk factor for staggered transfer times provided in this application. The values ​​in the figure show that when the bus arrival reliability is 0.72 and the ride-hailing arrival margin is 1 minute, the risk factor for staggered transfer times is 0.42, which exceeds the normal acceptable range. When the bus arrival reliability increases to 0.88 and the arrival margin increases to 3 minutes, the risk factor for staggered transfer times decreases to 0.16. When the arrival reliability reaches 0.94 and the arrival margin is not less than 4 minutes, the risk factor for staggered transfer times stabilizes below 0.08. It can be seen that the risk of staggered transfer times is not determined solely by the bus arrival deviation or the ride-hailing travel time, but is formed by the combined effect of bus arrival reliability, ride-hailing arrival margin, and the degree of exposure of connecting passengers. This application introduces this risk factor into the subsequent joint rolling optimization, which can avoid candidate solutions with high synergistic gains but high risk of transfer chain disruption from entering the target connecting scheme.

[0049] In step S5, the candidate connecting stops and connecting routes are jointly optimized by negative risk constraints through the collaborative gain degree and the transfer off-peak risk factor to obtain the target connecting stops and the target connecting routes.

[0050] In this embodiment, the joint rolling optimization of candidate connecting stops and connecting routes by applying negative risk constraints using the collaborative gain degree and the transfer off-peak risk factor to obtain the target connecting stop and target connecting route can be achieved through the following steps: Using candidate shuttle stops and shuttle routes as decision variables, the synergistic gain degree as a positive optimization driver, and the transfer off-peak risk factor as a negative risk constraint, a joint optimization space is constructed. Negative risk filtering is performed on the joint optimization space to eliminate candidate solutions whose transfer peak-shifting risk factors exceed the dynamic risk threshold, thereby obtaining a feasible solution domain that satisfies the negative risk constraint. Within the feasible solution domain, rolling optimization is performed with the cooperative gain degree as the main objective, the gain ranking of candidate solutions is updated, and the negative risk boundary is simultaneously verified. The candidate solution sequence of rolling optimization is balanced between gain and negative risk to obtain the target docking point and the target shuttle route.

[0051] It should be noted that, in this application, the joint optimization space refers to the search set of all candidate connection schemes, which are composed of positive gain indicators and negative risk indicators, taking candidate connection stops and connection routes as decision objects; the negative risk constraint refers to the constraint condition used to limit the risk upper limit of candidate connection schemes, based on the transfer off-peak risk factor; the feasible solution domain refers to the effective set of candidate connection schemes that meet the risk threshold requirements after negative risk filtering within the joint optimization space; the target connection stop refers to the optimal connection stop location obtained after joint rolling optimization; and the target connection route refers to the optimal connection route obtained after joint rolling optimization.

[0052] In practical implementation, firstly, each candidate shuttle stop is traversed sequentially. For each candidate shuttle stop, all corresponding shuttle routes are traversed sequentially, pairing each candidate shuttle stop with each shuttle route. Each pairing result is considered an independent candidate solution. The corresponding collaborative gain value and transfer off-peak risk factor value are simultaneously assigned to each candidate solution. The entire set of all assigned candidate solutions is considered the joint optimization space. Secondly, the total number of passengers waiting to be transferred in the target shuttle area within the shuttle response time window is counted. A baseline dynamic risk threshold of 0.3 is set. This baseline threshold is based on the general risk tolerance standard for urban shuttle scheduling, corresponding to a transfer chain disruption risk within 30% as a generally acceptable range. For every 20 additional passengers waiting to be transferred, the dynamic risk threshold is reduced by 0.05. The maximum dynamic risk threshold is... The lower limit is set to 0.1. This lower limit is used to avoid leaving no effective candidate solutions during peak passenger flow periods. Then, the transfer off-peak risk factor and dynamic risk threshold of each candidate solution in the joint optimization space are compared sequentially. Candidate solutions with transfer off-peak risk factor values ​​greater than the dynamic risk threshold are eliminated. The set of all remaining candidate solutions is taken as the feasible solution domain. Next, rolling optimization is performed in the feasible solution domain with the cooperative gain degree as the main objective. The gain ranking of candidate solutions is updated and the negative risk boundary is simultaneously verified. Finally, for each candidate solution in the updated gain ranking, the cooperative gain degree value is multiplied by 1 and the difference between the transfer off-peak risk factor value to calculate the equilibrium score value of each candidate solution. The candidate solution with the largest equilibrium score value is selected, and the candidate connecting stop corresponding to the candidate solution is taken as the target connecting stop, and the connecting travel path corresponding to the candidate solution is taken as the target connecting travel path.

[0053] In this embodiment, the following steps can be used to perform rolling optimization within the feasible solution domain with the cooperative gain degree as the primary objective, update the gain ranking of candidate solutions, and synchronously verify the negative risk boundary: Within the feasible solution domain, the cooperative gain degree is converted into the gain potential energy of the candidate solution, and the candidate solution is selected by rolling selection along the gain potential energy gradient direction to obtain the candidate solution sequence in the current time domain. For each solution in the candidate solution sequence, a negative risk boundary penetration determination is performed, and the candidate solutions are weighted by gain reduction based on the penetration depth to generate a weighted gain set; The gain weighted set is reintegrated into the rolling selection process within the feasible solution domain. The gain ranking of the previous rolling time domain is used as the warm start condition to generate an updated gain ranking and proceed to the next rolling time domain.

[0054] It should be noted that, in this application, the gain potential energy represents a quantitative value used to characterize the optimization potential of candidate solutions; the gain potential energy gradient direction represents the direction reflecting the trend of the high and low gain potential energy of candidate solutions, used to guide the direction of rolling optimization; the candidate solution sequence in the current time domain represents the ordered arrangement of candidate solutions in the current rolling time domain after rolling optimization along the gain potential energy gradient direction; the negative risk boundary penetration judgment represents the process of verifying whether the transfer and peak shifting risk factor of candidate solutions has broken through the negative risk boundary; the penetration depth represents the extent to which the transfer and peak shifting risk factor of candidate solutions exceeds the negative risk boundary, reflecting the degree of risk exceeding the limit; the gain weight reduction set represents the set of all candidate solutions after gain weight reduction processing of candidate solutions that have penetrated the negative risk boundary; the hot start condition represents the starting condition based on the gain ranking of the previous rolling time domain, used to improve the iteration efficiency of rolling optimization; and the updated gain ranking represents the new version of the candidate solution gain order arrangement result obtained after rolling iteration adjustment.

[0055] In specific implementation, firstly, the maximum and minimum values ​​of the collaborative gain of all candidate solutions within the feasible solution domain are extracted. Using the maximum value as the upper bound and the minimum value as the lower bound, the collaborative gain of each candidate solution is linearly scaled down to a uniform numerical range of 0 to 1. The corresponding calculated value is used as the gain potential energy of the candidate solution. A sliding window sorting method is adopted, setting a rolling time-domain window along the time axis from the beginning of the connection response time window. All candidate solutions within the feasible solution domain are sequentially arranged along the gradient direction of gain potential energy from high to low. The arrangement result of the candidate solutions within the coverage of the current rolling time-domain window is selected as the candidate solution sequence for the current time domain. Secondly, the difference between the transfer peak shifting risk factor and the negative risk boundary of each candidate solution in the candidate solution sequence is calculated one by one. Candidate solutions with positive difference calculation results are determined to have penetrated the negative risk boundary, and the corresponding positive value is used as the penetration depth of the candidate solution. Candidate solutions with non-positive difference calculation results are... If a solution is determined to have failed to penetrate the negative risk boundary, its penetration depth is recorded as 0. A proportional reduction method is used, with the ratio of the penetration depth of each candidate solution to the negative risk boundary value as the reduction ratio. The difference between 1 and the reduction ratio is used as the gain reduction coefficient for the corresponding candidate solution. The gain potential energy of each candidate solution is multiplied by its corresponding gain reduction coefficient to complete the gain reduction processing of all candidate solutions. All candidate solutions that have completed the weight reduction processing constitute the gain reduction set. Finally, the gain reduction set is reintegrated into the rolling optimization process within the feasible solution domain. The gain ranking obtained in the previous rolling time domain is used as the hot start condition for the iteration. Each candidate solution in the gain reduction set is inserted into the corresponding position of the gain ranking sequence in the previous rolling time domain according to the weighted gain potential energy value, and a new version of the candidate solution ranking order is formed. This ranking order is used as the updated gain ranking. At the same time, the rolling time domain window is slid forward by a preset step along the time axis to enter the optimization process of the next rolling time domain.

[0056] In step S6, the target shuttle stop location and target shuttle travel route are sent to the ride-hailing terminal and the bus dispatch terminal.

[0057] It should be noted that, in this application, the ride-hailing terminal refers to the ride-hailing vehicle terminal used to receive and execute connection dispatch instructions; the bus dispatch terminal refers to the bus dispatch system terminal used to receive and coordinate connection dispatch instructions.

[0058] In practice, the cloud dispatch platform encapsulates the target shuttle stop and the target shuttle route into a shuttle dispatch instruction message according to a preset message format. The target shuttle stop is encapsulated as the stop's latitude and longitude coordinates and stop number fields, and the target shuttle route is encapsulated as a path node sequence field arranged in the order of the route segments. Subsequently, the cloud dispatch platform establishes a data downlink to the ride-hailing terminal and the bus dispatch terminal through the mobile communication network. The encapsulated shuttle dispatch instruction message is sent to the ride-hailing terminal through this data downlink. The ride-hailing terminal parses the stop's latitude and longitude coordinates and the path node sequence and guides the ride-hailing vehicle to the target shuttle stop according to the target shuttle route on the in-vehicle navigation interface. At the same time, the shuttle dispatch instruction message is sent to the bus dispatch terminal through the data downlink. The bus dispatch terminal parses the stop number and matches the arrival and transfer information of the associated bus with the target shuttle stop to coordinate passenger transfers.

[0059] Therefore, this application demonstrates that the target shuttle stop and target shuttle route can be distributed to ride-hailing terminals and bus dispatch terminals. Firstly, by acquiring the road network traffic status of the target shuttle area and the estimated arrival times of associated bus routes, and determining the shuttle response time window accordingly, subsequent dispatching can be established within the time frame jointly defined by road traffic changes and bus arrival processes, preventing shuttle arrangements from deviating from the actual transfer sequence. Secondly, by determining the shuttle priority based on the order location of passengers awaiting shuttle access, road network traffic status, and the spatiotemporal convergence of passenger flow at candidate shuttle stops within the shuttle response time window, shuttle priority can be determined, balancing passenger convergence needs and road accessibility, filtering out stops with insufficient passenger capacity or poor traffic conditions, and reducing passenger dispersion, vehicle detours, and extended waiting times caused by fixed-point dispatching. Furthermore, the shuttle priority is determined by the toll cost of the shuttle route and the shuttle priority of candidate shuttle stops. The collaborative gain factor incorporates the passenger carrying value of a stop and the transit cost of a ride-hailing vehicle to that stop into the same evaluation process, avoiding the separation between stop selection and vehicle route planning. It also assesses the risk of peak-hour transfers by combining the arrival reliability of bus services and the arrival margin of ride-hailing routes, identifying candidate combinations with low transit costs but mismatched transfer timings. Finally, through the collaborative gain factor and peak-hour transfer risk factors, a joint rolling optimization of candidate stop locations and routes is performed with negative risk constraints. This allows for synchronized adjustment of stop locations and routes based on changes in road network traffic conditions, ride-hailing vehicle locations, and bus arrival status. The scheduling results are then distributed to ride-hailing terminals and bus dispatch terminals, reducing route mismatches after stop adjustments and transfer timing mismatches after route changes. This, in turn, reduces vehicle detour distances and passenger waiting times, improving the overall operational efficiency of the ride-hailing and bus connection process.

[0060] In summary, the technical solution adopted in this application can achieve collaborative optimization of connection points and travel routes in dynamic connection scenarios, thereby improving collaborative connection efficiency.

[0061] Example 2 This application provides a ride-hailing and public transport collaborative dispatch system based on a cloud dispatch platform, referencing... Figure 4 As shown in the figure, this is a modular structure diagram of a ride-hailing and public transport collaborative dispatch system based on a cloud dispatch platform according to this embodiment of the application. The collaborative dispatch system includes: The connection response time window generation module 100 is used by the cloud scheduling platform to obtain the road network traffic status of the target connection area and the estimated arrival time of the associated bus routes, and to determine the connection response time window through the estimated arrival time; The candidate shuttle stop evaluation module 200 is used to filter candidate shuttle stops based on the order location of the passengers to be picked up and the road network traffic status, and to determine the shuttle priority of the candidate shuttle stops by the spatiotemporal convergence of passenger flow within the shuttle response time window. The connecting path coordination gain module 300 is used to generate connecting routes based on the real-time location of the ride-hailing vehicle and candidate connecting stops, and to determine the coordination gain between candidate connecting stops and connecting routes by the toll cost of the connecting routes and the connecting priority of candidate connecting stops. The Transfer Peak Shift Risk Assessment Module 400 is used to assess the peak shift shift risk by associating the arrival reliability of bus routes with the arrival margin of ride-hailing connection routes, and to obtain the transfer peak shift risk factor. The joint rolling optimization decision module 500 is used to perform joint rolling optimization of candidate connecting stops and connecting routes by applying negative risk constraints to the collaborative gain degree and the transfer off-peak risk factor, so as to obtain the target connecting stop and the target connecting route. The collaborative scheduling scheme distribution module 600 is used to distribute the target shuttle stop and target shuttle travel route to the ride-hailing terminal and the bus dispatch terminal.

[0062] It should also be noted that the reference Figure 5 As shown, Figure 5This diagram illustrates an application scenario of ride-hailing and public transportation collaborative connection under the cloud dispatch platform provided in this application. The cloud dispatch platform, as the core processing unit, internally includes a data management module, a collaborative dispatch module, and an operation monitoring module. These modules are used to receive passenger order locations submitted by user terminals, obtain estimated arrival times of public transportation vehicles, bus stop information, real-time locations of ride-hailing vehicles, and road network traffic status. User terminals are used by passengers to initiate connection requests and receive target connection stops and connection prompts. Management terminals are used by dispatchers to view the platform's operational status and connection execution. A communication network is used to implement the cloud dispatch platform. The platform facilitates data interaction between the platform and user terminals, management terminals, and the travel service layer. The travel service layer includes ride-hailing vehicles, buses, and bus stops. The cloud dispatch platform determines the connection response time window based on the estimated arrival time of buses, filters candidate connection stops based on passenger order locations, and generates connection routes by combining the real-time location of ride-hailing vehicles. Furthermore, the platform distributes the target connection stops and target connection routes to ride-hailing terminals and bus dispatch terminals through collaborative dispatch results, enabling ride-hailing vehicles to reach the vicinity of bus stops according to the target routes to complete the connection and transfer, thereby realizing the linkage dispatch between ride-hailing vehicles and buses.

[0063] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and 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 process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0064] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0065] It should also be noted that 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 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.

Claims

1. A method for coordinating and scheduling online car-hailing and public transportation based on a cloud scheduling platform, characterized in that, Includes the following steps: The cloud dispatch platform obtains the road network traffic status of the target connection area and the estimated arrival time of the associated bus routes, and determines the connection response time window through the estimated arrival time; Candidate shuttle stops are selected based on the order location of passengers to be picked up and the road network traffic status, and the pick-up priority of the candidate shuttle stops is determined by the spatiotemporal convergence of passenger flow within the pick-up response time window. A connecting route is generated based on the real-time location of the ride-hailing vehicle and candidate pick-up stops. The collaborative gain between the candidate pick-up stops and the connecting route is determined by the toll cost of the connecting route and the pick-up priority of the candidate pick-up stops. By associating the arrival reliability of bus routes with the arrival margin of ride-hailing connection routes, the risk of staggered transfer times is assessed, and the staggered transfer risk factor is obtained. By combining the synergistic gain and the transfer off-peak risk factor, the candidate connecting stops and connecting routes are jointly optimized under negative risk constraints to obtain the target connecting stops and target connecting routes. The target shuttle stop and target shuttle route are sent to the ride-hailing terminal and the bus dispatch terminal.

2. The method for coordinated dispatching of ride-hailing and public transportation based on a cloud dispatching platform as described in claim 1, characterized in that, Determining the connection response time window based on the estimated arrival time specifically includes: Obtain the estimated arrival time of the associated bus routes at the corresponding bus stops within the target connection area; Based on the estimated arrival time, extract the base arrival time of the bus service entering the target connection area; Based on the aforementioned arrival reference time, a buffer zone for advance pick-up of ride-hailing vehicles and a permissible waiting period for passenger transfers are set. A connection response time window is generated based on the advance connection buffer segment and the transfer waiting allowance segment.

3. The method for coordinated dispatching of ride-hailing vehicles and public transportation based on a cloud dispatching platform as described in claim 1, characterized in that, The selection of candidate shuttle stops based on the order location of the passengers to be picked up and the road network traffic status specifically includes: Extract spatial clusters of orders within the target pick-up area based on the order locations of passengers to be picked up; The initial set of shuttle stops is determined based on the matching relationship between the order space clusters and the associated bus stops in terms of their connection radius. Based on the road network traffic status, the initial set of connecting stops is checked for road segment traffic constraints to obtain a set of reachable stops that meet the connecting conditions. Candidate shuttle stops are selected based on the temporary parking capacity conditions of each stop in the set of reachable stops and the passenger pedestrian access boundary.

4. The method for coordinated dispatching of ride-hailing vehicles and public transportation based on a cloud dispatching platform as described in claim 1, characterized in that, The priority of candidate shuttle stops is determined by the spatiotemporal convergence of passenger flow within the shuttle response time window. This includes: Extract the arrival sequence of pending shuttle orders corresponding to the candidate shuttle stops based on the shuttle response time window; Based on the arrival sequence of the pending shuttle orders, the concentrated arrival segments of the orders are identified, thereby determining the passenger flow convergence period characteristics of the candidate shuttle stops; Based on the spatial distribution relationship between the order locations of passengers to be transferred and the candidate transfer stops, the spatial aggregation features of passenger flow at the candidate transfer stops are extracted; Based on the passenger flow convergence time period characteristics and the passenger flow spatial convergence characteristics, the spatiotemporal convergence degree of passenger flow at the candidate shuttle stop within the shuttle response time window is determined; The candidate shuttle stops are prioritized by the spatiotemporal convergence of passenger flow to obtain the shuttle priority of the candidate shuttle stops.

5. The method for coordinated dispatching of ride-hailing vehicles and public transportation based on a cloud dispatching platform as described in claim 1, characterized in that, The specific steps for generating a shuttle route based on the real-time location of the ride-hailing vehicle and candidate pick-up points include: Obtain the real-time location of ride-hailing vehicles within the target pick-up area; Based on the real-time location and each candidate pick-up stop, establish the pick-up route pointing relationship from the ride-hailing vehicle to the candidate pick-up stop; Based on the connection path pointing relationship and the road network traffic status, a sequence of passable road segments that meet the connection response time window requirements is selected. Based on the sequence of passable road segments, a connecting route is generated for ride-hailing vehicles to reach candidate pick-up points.

6. The method for coordinated dispatching of ride-hailing vehicles and public transportation based on a cloud dispatching platform as described in claim 1, characterized in that, The coordination gain between candidate shuttle stops and shuttle routes is determined by the toll cost of the shuttle route and the shuttle priority of the candidate shuttle stops. Specifically, this includes: Determine the toll cost of connecting routes; Extract the connection revenue characteristics of candidate shuttle stops based on their connection priority; Based on the passage cost and the connection benefit characteristics, a collaborative gain analysis is performed on the candidate connection stops and connection routes to obtain the collaborative gain degree between the candidate connection stops and connection routes.

7. The method for coordinated dispatching of ride-hailing vehicles and public transportation based on a cloud dispatching platform as described in claim 1, characterized in that, By assessing the peak-hour risk of transfer connections through the correlation between the arrival reliability of bus routes and the arrival margin of ride-hailing connection routes, the specific peak-hour transfer risk factors include: Determine the arrival reliability of associated bus routes; The arrival time confidence region of ride-hailing vehicles is determined based on the arrival margin of the ride-hailing connection route; Based on the arrival reliability of associated bus routes and the arrival time confidence region of ride-hailing vehicles, a time connection conflict analysis is performed under the condition of link breakage triggering to obtain the conditional risk field of transfer connection link breakage. The passenger flow at candidate connecting stops is assessed for off-peak risk using the conditional risk field to obtain the off-peak transfer risk factor.

8. The method for coordinated dispatching of ride-hailing vehicles and public transportation based on a cloud dispatching platform as described in claim 1, characterized in that, By jointly performing rolling optimization with negative risk constraints on candidate connecting stops and connecting routes using the synergistic gain degree and the transfer off-peak risk factor, the target connecting stops and target connecting routes are obtained, specifically including: Using candidate shuttle stops and shuttle routes as decision variables, the synergistic gain degree as a positive optimization driver, and the transfer off-peak risk factor as a negative risk constraint, a joint optimization space is constructed. Negative risk filtering is performed on the joint optimization space to eliminate candidate solutions whose transfer peak-shifting risk factors exceed the dynamic risk threshold, thereby obtaining a feasible solution domain that satisfies the negative risk constraint. Within the feasible solution domain, rolling optimization is performed with the cooperative gain degree as the main objective, the gain ranking of candidate solutions is updated, and the negative risk boundary is simultaneously verified. The candidate solution sequence of rolling optimization is balanced between gain and negative risk to obtain the target docking point and the target shuttle route.

9. A method for coordinated dispatching of ride-hailing vehicles and public transportation based on a cloud dispatching platform as described in claim 8, characterized in that, Within the feasible solution domain, rolling optimization is performed with the cooperative gain as the primary objective, updating the gain ranking of candidate solutions and synchronously verifying the negative risk boundary. Specifically, this includes: Within the feasible solution domain, the cooperative gain degree is converted into the gain potential energy of the candidate solutions, and the candidate solutions are selected by rolling selection along the gain potential energy gradient direction to obtain the candidate solution sequence in the current time domain. For each solution in the candidate solution sequence, a negative risk boundary penetration determination is performed, and the candidate solutions are weighted by gain reduction based on the penetration depth to generate a weighted gain set; The gain weighted set is reintegrated into the rolling selection process within the feasible solution domain. The gain ranking of the previous rolling time domain is used as the warm start condition to generate an updated gain ranking and proceed to the next rolling time domain.

10. A ride-hailing and public transport collaborative dispatching system based on a cloud dispatching platform, used to execute the ride-hailing and public transport collaborative dispatching method based on a cloud dispatching platform as described in any one of claims 1 to 9, characterized in that, The collaborative scheduling system includes: The connection response time window generation module is used by the cloud scheduling platform to obtain the road network traffic status of the target connection area and the estimated arrival time of the associated bus routes, and to determine the connection response time window through the estimated arrival time; The candidate shuttle stop evaluation module is used to filter candidate shuttle stops based on the order location of passengers to be picked up and the road network traffic status, and to determine the shuttle priority of candidate shuttle stops by the spatiotemporal convergence of passenger flow within the shuttle response time window; The connecting path coordination gain module is used to generate connecting routes based on the real-time location of ride-hailing vehicles and candidate connecting stops, and to determine the coordination gain between candidate connecting stops and connecting routes by the toll cost of the connecting routes and the connecting priority of candidate connecting stops. The transfer off-peak risk assessment module is used to assess the off-peak risk of transfer connection sequence by associating the arrival reliability of bus routes and the arrival margin of ride-hailing connection routes, and obtain the transfer off-peak risk factor. The joint rolling optimization decision module is used to jointly roll optimize the candidate connecting stops and connecting routes by applying negative risk constraints to the collaborative gain degree and the transfer off-peak risk factor, so as to obtain the target connecting stops and the target connecting routes. The collaborative scheduling scheme distribution module is used to distribute the target shuttle stop points and target shuttle routes to ride-hailing terminals and bus dispatch terminals.

Images