A flexible public transport travel demand identification method and system

Abstract

This invention discloses a method and system for identifying flexible public transport travel demand, comprising the following steps: S1, acquiring multi-source heterogeneous travel data, cleaning and spatiotemporally aligning the acquired data, and extracting the origin-destination (OD) and spatiotemporal trajectory features of individuals; S2, based on the travel OD features and spatiotemporal trajectory features, using an adaptive density spatiotemporal clustering algorithm to identify demand hotspot clusters with high potential for flexible public transport travel, and dynamically dividing the service area of ​​flexible public transport according to the distribution boundary and timeliness of the demand clusters; S3, quantifying the demand features within the service area, combining real-time traffic conditions and environmental factors, using a nonparametric Bayesian model to predict the implicit demand within future time windows, classifying the demand into several levels based on the prediction results, and triggering corresponding response mechanisms. This invention's proactive demand identification: by integrating multi-source data such as mobile phone signaling and IC cards, combined with spatiotemporal clustering algorithms, it can mine micro-level potential travel demands from macro-level massive data, transforming "passive order acceptance" into "proactive discovery," effectively improving the lead time and accuracy of demand identification.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and public transportation operation management technology, specifically to a flexible public transportation travel demand identification method and system that combines multi-source data mining, dynamic regional planning and game theory scheduling. Background Technology

[0002] With the acceleration of urbanization and the increasing diversification of residents' travel needs, the traditional public transportation model with fixed routes, fixed schedules, and fixed stops can no longer fully meet the public's travel demands, especially in urban fringe areas, new districts, and the "last mile" areas, where demand is dispersed and exhibits significant tidal characteristics. To address this issue, new public transportation models such as responsive buses, flexible buses, and customized buses have emerged.

[0003] However, existing flexible public transport technologies still have the following major drawbacks in practical applications:

[0004] First, demand identification is passive and delayed. Most existing technologies rely on passengers proactively initiating reservations or placing orders in real time through mobile apps. The system can only respond to explicit orders and lacks the ability to proactively discover and predict potential travel demand. This leads to an inefficient situation of "cars waiting for orders" or "orders waiting for cars," especially in the initial demand phase, where it is easy to fall into a vicious cycle of passenger loss due to lack of car response and vehicle withdrawal due to lack of orders.

[0005] Second, the service area planning is rigid. Traditional customized bus routes or service areas are often determined by human experience and remain fixed for a long period of time. This static planning cannot adapt to the dynamic shifts in travel hotspots over time (such as morning and evening rush hours) and with events (such as large-scale events and weather changes), resulting in a mismatch between transport capacity resources and spatial distribution of demand.

[0006] Third, scheduling and route planning lack coordination. Existing technologies mostly use single vehicle routing problem (VRP) heuristic algorithms for solving problems, often focusing only on a single objective (such as the shortest path) without fully considering the game relationship between operator revenue and passenger service experience under dynamically changing demands. It is difficult to achieve the optimal balance between reducing operating costs and improving passenger satisfaction. Summary of the Invention

[0007] A flexible public transport demand identification method and system is proposed to address the problems of passive demand identification, inflexible service area planning, and poor scheduling coordination in existing technologies.

[0008] A method for identifying flexible public transport travel demand includes the following steps:

[0009] S1. Acquire multi-source heterogeneous travel data, clean and spatiotemporally align the acquired data, and extract the origin and destination (OD) and spatiotemporal trajectory features of individuals.

[0010] S2. Based on the travel OD characteristics and spatiotemporal trajectory characteristics, an adaptive density spatiotemporal clustering algorithm is used to identify demand hotspot clusters with high travel potential for flexible public transportation, and the service area of ​​flexible public transportation is dynamically divided according to the distribution boundary and timeliness of the demand clusters.

[0011] S3. Quantify the demand within the service area by feature, combine real-time traffic conditions and environmental factors, use a non-parametric Bayesian model to predict the implicit demand within the future time window, divide the demand into several levels according to the prediction results, and trigger the corresponding response mechanism.

[0012] S4. Based on the demand level and real-time vehicle status, construct a dynamic game model with passenger time cost and vehicle operating cost as the objectives, solve the vehicle route planning and station stopping scheme, and generate collaborative scheduling instructions.

[0013] S5. Collect passenger feedback data after actual operation, and reverse correct the parameters of the spatiotemporal clustering algorithm and the prior distribution of the nonparametric Bayesian model to achieve self-learning iteration of demand identification accuracy.

[0014] Furthermore, in step S2, the adaptive density spatiotemporal clustering algorithm is an improved DBSCAN algorithm that introduces a time decay factor and a spatial reachability threshold. This algorithm calculates the spatiotemporal distance between sample points. The formula is:

[0015]

[0016] in, This refers to Euclidean distance or road network distance. For the time difference, α and β are the weighting coefficients for the spatial and temporal dimensions, respectively, and these weighting coefficients are dynamically adjusted according to the land use characteristics of the service area.

[0017] Only when When the demand density in a given area is less than a set threshold and the demand density in that area reaches a preset density threshold, that area is designated as the current service sub-area for flexible public transport.

[0018] Further, in step S3, the response mechanism includes a high-frequency response mechanism, an on-demand response mechanism, and an intensive response mechanism; the specific steps for triggering the corresponding response mechanism include: if the predicted demand density is higher than a first preset threshold and the time urgency is higher than a preset standard, the high-frequency response mechanism is triggered, and an operation mode combining fixed routes and flexible stations is adopted; if the predicted demand density is between the first preset threshold and a second preset threshold, the on-demand response mechanism is triggered, and the vehicle route is completely determined by real-time dynamic requests, and a maximum detour coefficient constraint is set; if the predicted demand density is lower than the second preset threshold, the intensive response mechanism is triggered, and the real-time response service in the area is suspended, switching to accepting only reservation services or guiding passengers to nearby hub stations.

[0019] Furthermore, in step S4, the dynamic game model uses the vehicle as one party and the passenger set as the other party; it constructs a system total revenue function that includes passenger waiting time penalty, in-vehicle time penalty, and vehicle mileage cost; and by solving the Nash equilibrium, it obtains a vehicle route planning scheme and a passenger boarding point recommendation scheme that maximizes the system total revenue.

[0020] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method.

[0021] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0023] 1. Proactive Demand Identification: By integrating multi-source data such as mobile phone signaling and IC cards, and combining spatiotemporal clustering algorithms, it is possible to mine micro-level potential travel demands from macro-level massive data, transforming "passive order acceptance" into "proactive discovery," effectively improving the lead time and accuracy of demand identification.

[0024] 2. Dynamic Service Area Planning: Utilizing the improved DBSCAN algorithm, the service area is no longer fixed, but dynamically expands and deforms in real time as demand hotspots shift, greatly improving the utilization efficiency of vehicle resources and the accuracy of service coverage.

[0025] 3. Tiered Response and Coordinated Scheduling: Demand is tiered using nonparametric Bayesian prediction, triggering different operational modes such as high-frequency, on-demand, or intensive scheduling for different tiers. A dynamic game theory model is introduced for scheduling, seeking the optimal solution between passenger waiting time, in-vehicle comfort, and enterprise operating costs, thereby maximizing the overall efficiency of the system. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart of the flexible public transportation travel demand identification method provided in an embodiment of the present invention.

[0028] Figure 2 This is a schematic diagram of dynamic service area planning in an embodiment of the present invention.

[0029] Figure 3 This is a logical block diagram of collaborative scheduling and game theory solution in an embodiment of the present invention. Detailed Implementation

[0030] To make the technical means, creative features, objectives and effects of the present invention easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0031] Example 1

[0032] like Figure 1 As shown, this embodiment provides a flexible public transportation travel demand identification method, which can run on a cloud server or a traffic dispatch center, and specifically includes the following steps:

[0033] Step S1: Multi-source data perception and fusion.

[0034] The system accesses multi-source data in real time through API interfaces, including: mobile phone signaling data (reflecting the real-time location and movement of a large population), public transport IC card and QR code scanning data (reflecting public transport preferences), ride-hailing historical order data (reflecting high-value travel intentions), and road traffic flow data.

[0035] The data processing module cleans the aforementioned multi-source heterogeneous data, removing noise and outliers; it uses a map matching algorithm to map GPS trajectories onto the road network topology; it uses a station identification algorithm to infer an individual's residence and workplace, and combines spatiotemporal trajectories to extract origin-destination (OD) features to construct a city-level travel demand map.

[0036] Step S2: Potential demand mining and dynamic planning of service areas.

[0037] Based on the extracted OD features, this invention employs an improved DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm for spatiotemporal clustering. Traditional DBSCAN only considers spatial distance; this embodiment introduces a temporal dimension, defining spatiotemporal distance.

[0038] .

[0039] For example, in residential areas, flexible public transport requests generated between 7:30 and 8:30 AM exhibit high temporal aggregation, and the time weight β can be appropriately increased in this case. The algorithm calculates the density of all potential travel points within the area. If the point density of a certain area is greater than the threshold MinPts and the points are connected, it is identified as a "demand hotspot cluster".

[0040] Based on the convex hull boundary of the demand hotspot clusters, the system dynamically generates "flexible public transport service areas." For example... Figure 2 As shown, the area has an irregular shape, closely matches the actual road network and demand distribution, and is updated by sliding with time windows (such as every 15 minutes) to achieve dynamic drift of the service area.

[0041] Step S3: Demand grading forecasting and response mechanism triggering.

[0042] Within the designated service area, the system uses a nonparametric Bayesian model (such as the Hierarchical Dirichlet Process, HDP) to make short-term predictions of implicit demand. This model does not pre-determine the distribution pattern and can automatically adjust according to the sparsity of historical data, making it suitable for handling the characteristics of flexible public transport demand with large fluctuations.

[0043] Based on the prediction results, the system divides the current demand status of the region into three levels and triggers the corresponding operation mode design:

[0044] * Level 1 (High-Frequency Response): When the predicted demand density is extremely high (such as during peak hours), a "bus-like" mode is triggered. The system generates a virtual fixed trunk line, and vehicles run on the trunk line, but are allowed to flexibly stop at stations within a certain range (such as 200 meters) based on passenger locations.

[0045] * Level 2 (On-Demand Response): When the predicted demand density is moderate, the "dynamic carpooling" mode is triggered. The vehicle route is entirely determined by real-time dynamic requests. The algorithm calculates the optimal order of accepting orders, but strictly limits the vehicle detour coefficient (e.g., the detour does not exceed 1.3 times the straight-line distance) to ensure efficiency.

[0046] * Level 3 (Intensive Response): When the predicted demand density is extremely low (such as during off-peak hours), the "Reservation / Guidance" mode is triggered, suspending real-time cruising and switching to accumulating a certain number of orders before departure, or recommending passengers use shared bicycles to connect to the nearest regular bus stop.

[0047] Step S4: Coordinated scheduling and route planning.

[0048] For flexible site scheduling in Level 2 or Level 1, this invention constructs a collaborative scheduling model based on dynamic game theory (see [link]). Figure 3 ).

[0049] * Game players: Party A is the bus operator (vehicle), and Party B is the passenger.

[0050] * Strategy Space: Passenger strategy is to place an order or cancel, while vehicle strategy is to accept the order and select a route.

[0051] * Revenue Function: Construct the system's total revenue function.

[0052] .in, and These refer to passenger waiting time and time spent inside the vehicle, respectively. This refers to the vehicle's mileage.

[0053] * Solution Process: An improved genetic simulated annealing algorithm is used to find the Nash equilibrium point of the model. During the iteration process, the algorithm continuously attempts to insert new orders. The vehicle route planning is only updated when the addition of a new order increases the total system revenue (i.e., the increase in passenger time cost is less than the implicit benefits from savings in vehicle operating costs or improved passenger experience). This generated route planning ensures that passenger waiting times are within an acceptable range while controlling operating costs.

[0054] Step S5: Feedback Iteration and Model Self-Learning.

[0055] After a vehicle completes its operational task, the system collects feedback information such as actual passenger reviews and actual disembarkation deviation data. This data is used as a supervisory signal to update the cluster density parameter in step S2 and the Bayesian prior distribution in step S3. For example, if a region frequently experiences predicted demand but low actual order rates, the weight of the land use type in that region is reduced, thereby gradually improving the model's adaptability to the characteristics of specific regions.

[0056] Example 2

[0057] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of the method described in Embodiment 1.

[0058] Example 3

[0059] This embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0060] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method and system for identifying flexible public transport travel demand, characterized in that... Includes the following steps: S1. Acquire multi-source heterogeneous travel data, clean and spatiotemporally align the acquired data, and extract the origin and destination (OD) and spatiotemporal trajectory features of individuals. S2. Based on the travel OD characteristics and spatiotemporal trajectory characteristics, an adaptive density spatiotemporal clustering algorithm is used to identify demand hotspot clusters with high travel potential for flexible public transportation, and the service area of ​​flexible public transportation is dynamically divided according to the distribution boundary and timeliness of the demand clusters. S3. Quantify the demand within the service area by feature, combine real-time traffic conditions and environmental factors, use a non-parametric Bayesian model to predict the implicit demand within the future time window, divide the demand into several levels according to the prediction results, and trigger the corresponding response mechanism. S4. Based on the demand level and real-time vehicle status, construct a dynamic game model with passenger time cost and vehicle operating cost as the objectives, solve the vehicle route planning and station stopping scheme, and generate collaborative scheduling instructions. S5. Collect passenger feedback data after actual operation, and reverse correct the parameters of the spatiotemporal clustering algorithm and the prior distribution of the nonparametric Bayesian model to achieve self-learning iteration of demand identification accuracy. Furthermore, in step S2, the adaptive density spatiotemporal clustering algorithm is an improved DBSCAN algorithm that incorporates a time decay factor and a spatial reachability threshold. This algorithm calculates the spatiotemporal distance between sample points. The formula is: in, This refers to Euclidean distance or road network distance. For the time difference, α and β are the weighting coefficients for the spatial and temporal dimensions, respectively, and these weighting coefficients are dynamically adjusted according to the land use characteristics of the service area. Only when When the demand density in a given area is less than a set threshold and the demand density in that area reaches a preset density threshold, that area is designated as the current service sub-area for flexible public transport.

2. The flexible public transport travel demand identification method and system according to claim 1, characterized in that... In step S3, the response mechanism includes a high-frequency response mechanism, an on-demand response mechanism, and a centralized response mechanism. The specific steps for triggering the corresponding response mechanism include: if the predicted demand density is higher than a first preset threshold and the time urgency is higher than a preset standard, the high-frequency response mechanism is triggered, and an operation mode combining fixed routes and flexible stations is adopted; if the predicted demand density is between the first preset threshold and a second preset threshold, the on-demand response mechanism is triggered, and the vehicle route is completely determined by real-time dynamic requests, with a maximum detour coefficient constraint set; if the predicted demand density is lower than the second preset threshold, the centralized response mechanism is triggered, and the real-time response service in the area is suspended, switching to accepting only reservation services or guiding passengers to nearby hub stations.

3. The flexible public transport travel demand identification method and system according to claim 1, characterized in that... In step S4, the dynamic game model uses the vehicle as one side of the game and the passenger set as the other side. A system total revenue function is constructed, which includes passenger waiting time penalty, in-vehicle time penalty, and vehicle mileage cost. By solving the Nash equilibrium, a vehicle route planning scheme and a passenger boarding point recommendation scheme that maximize the system total revenue are obtained.

Images