A fixed station and variable line customized bus multi-level station design method

Abstract

The application discloses a kind of fixed station not fixed line custom bus multi-level site design method, innovatively proposes a kind of multi-level site design method, can fully consider system design service parameter, overcome the problem that traditional clustering method is prone to produce too large demand cluster and site is too sparse.Proceed and algorithm of site design can effectively control site service radius;The multi-level site design method proposed first generates main site based on demand density clustering, and determines the number of required sub-station by buffer zone analysis to main site demand, then the main site demand is divided twice, finally determines the position of main site and sub-station based on actual road network, can effectively solve the practical problems of site sparse and long passenger walking distance caused by larger cluster cluster in existing design method.

Description

Technical Field

[0001] This invention relates to a multi-level station design method for customized buses with fixed stops but not fixed routes, belonging to the interdisciplinary application field of big data information mining and public transportation system design. Background Technology

[0002] As cities grow and residents experience a faster pace of life and significantly improved living standards, their demands for convenient, comfortable, and fast travel are constantly increasing. Customized bus services, relying on internet technology and guided by passenger needs, plan vehicle routes based on actual passenger travel demand distribution and travel time requirements. Compared to traditional buses, this better meets passengers' personalized travel needs, enhances their travel experience, improves travel efficiency, and effectively absorbs private car travel demand, reducing the number of motor vehicles on the road and alleviating traffic congestion and air pollution.

[0003] Customized bus systems provide passengers with a more effective and safer space for epidemic prevention, avoiding unnecessary close contact with other travelers whose journeys are not the same, greatly reducing the risk of cross-infection. The advantages of traceable user travel routes and ease of tracking have enabled them to play a significant role in public transportation in the post-pandemic era. Fixed stops but flexible routes are a common operating model in customized bus systems. Buses provide travel services to passengers within a certain area, with fixed stops but flexible routes. Passengers select their boarding and alighting stops and travel time from a database of stops on a mobile app, and the system dispatches buses and plans the optimal route in real time based on actual demand. This model is currently widely used in cities such as Beijing and Xi'an.

[0004] Selecting appropriate station locations to serve and accommodate more travelers and their travel needs is a crucial factor in the efficient operation of a fixed-stop, variable-route customized bus system. A well-planned station layout can effectively reduce passenger walking distances, maintain high occupancy rates, and increase the service level and profitability of the customized bus system. Therefore, exploring a scientific method for designing fixed-stop, variable-route customized bus stations has significant practical implications.

[0005] In existing technologies, Ma Xiaolei et al. disclosed a customized bus stop location selection method based on ride-hailing data (CN 109583611 A). This method extracts travel origin-destination (OD) data from ride-hailing order data, supplements the passenger travel demand collection method, and simplifies ride-hailing data into the concepts of inter-regional and intra-regional flow, thereby improving the accuracy of stop location selection. However, it has the following drawbacks: it cannot use customized bus service design parameters (such as walking distance) as control variables for stop division; the clustering algorithm used is susceptible to outliers and noise points; the clustering algorithm parameter K is determined based on empirical values, which can easily generate excessively large clusters and traffic zones, failing to guarantee that passenger walking distances remain within a certain range.

[0006] Zhang et al. disclosed a customized bus stop layout method based on mobile phone signaling data (CN 112150796A). This method identifies the arrival and departure populations and origin-destination (OD) of hub stations using daily mobile phone signaling data and employs the DBSCAN clustering algorithm to determine the location of ride-sharing stations. However, it has the following drawbacks: the origin and destination locations of the demand data are not precise; and the clustering algorithm tends to produce large clusters, leading to a situation where stations are relatively sparse in areas with high demand density.

[0007] Deficiencies of existing methods or inventions:

[0008] 1) Currently, there is no publicly available station design method directly applicable to the fixed-stop, variable-route customized bus system;

[0009] 2) The service design parameters of the customized bus system cannot be fully considered;

[0010] 3) The calculation of the number of stations corresponding to the demand clusters lacks scientific rigor. The algorithm used cannot comprehensively consider the station service radius limitation and the distribution of demand density, which can easily lead to excessively large demand clusters, resulting in the actual problems of overly sparse stations and excessively long walking distances for passengers. Summary of the Invention

[0011] Based on the shortcomings and deficiencies of the aforementioned customized bus stop site selection methods, this invention constructs a multi-level stop design method for fixed-stop, variable-route customized buses. This method fully considers the travel system design parameters and innovatively proposes a multi-level stop design approach that comprehensively considers the station service radius constraints and demand density distribution, overcoming the problems of excessively large demand clusters and overly sparse stations that are easily generated by traditional clustering methods. The proposed stop design method improves the scientific rigor of dynamic customized bus stop design with fixed stops and variable routes, and this method has universal applicability in different cities.

[0012] To achieve the above objectives, the present invention adopts the following technical solution:

[0013] A method for designing multi-level bus stops for customized buses with fixed stops but variable routes, comprising the following steps:

[0014] Step 1: Data Collection: Collect potential demand data, such as ride-hailing platform order data, public transport IC card data, etc. Fields that must be included or estimated in travel big data include the origin location, destination location, and departure time timestamp; extract vector geographic data of the target area as a map dataset;

[0015] Step 2: Extracting travel demand information: Extracting or calculating travel (order) demand information, including origin and destination coordinates, departure time, etc.

[0016] Step 3: Main Site Division:

[0017] S1: Spatial clustering of travel demand origin and destination: Based on the origin and destination longitude and dimension coordinates of the demand, the density-based noise application spatial clustering algorithm (DBSCAN) is adopted. The spatial clustering is performed with the service radius R of the design station and the number of demands minPts corresponding to the minimum occupancy rate as parameters to generate the origin cluster and destination cluster of the demand.

[0018] S2: Calculate the spatial coordinates of the main site: Using the K-means clustering algorithm, the starting and ending points of the demand in each cluster are used as samples. The number of clusters K is set to 1. Clustering is used to obtain the demand clusters. The centroid coordinates of each cluster are the coordinates of the main site.

[0019] Step 4: Subsite division:

[0020] S1: Calculate the service area of ​​each main site based on buffer analysis: Generate circular buffers as demand buffers for the demand start and end points in the demand cluster, with the radius set as the site service radius R, and merge the demand buffers in the same cluster to obtain the service area of ​​each main site;

[0021] S2: Calculate the number of sub-sites: Calculate the service area A of each main site, and calculate the number of sub-sites Q to be divided based on the following formula:

[0022]

[0023] S3: Sub-site demand re-clustering: The K-means clustering algorithm is used with the start and end points of demand in each cluster as samples, and the number of clusters is set to Q. The sub-demand clusters of each main site are obtained by clustering, and the centroid of the sub-site demand clusters is obtained.

[0024] Step 5: Deduplicating sites:

[0025] S1: Analyze the spatial distribution of the boarding and alighting sub-stations generated at the starting and ending points of the demand; calculate the distance between each boarding sub-station and each alighting sub-station, and mark the stations with a station spacing less than the maximum service radius R as stations to be adjusted;

[0026] S2: Merge the sites to be adjusted using a center coordinate calculation method based on demand weights.

[0027] Step 6: Calculate the location of the sub-sites: Project the deduplicated sub-sites onto the road network to obtain the actual site location coordinates.

[0028] Preferably, the density-based noise applied spatial clustering (DBSCAN) used in the third step is an unsupervised machine learning clustering algorithm that can cluster clusters of arbitrary shapes, is insensitive to outliers, and does not require a pre-defined number of clusters.

[0029] Preferably, the fusion of demand buffers in the fourth step is the process of merging intersecting buffer elements from the demand-generated buffer elements of the same demand cluster into a single buffer surface element.

[0030] Preferably, the cluster buffer analysis in the fourth step can calculate the service area of ​​the site under the premise of considering the walking distance constraint, and finally calculate the number of sub-sites Q that each cluster needs to be further divided into as the input parameter for the next Kmeans clustering, which makes up for the defect of the traditional algorithm that subjectively determines the number of sub-sites based on the designer's experience.

[0031] Preferably, in the fifth step, a weighted site coordinate calculation method is used to merge sites that are too close together. Assuming the coordinates of two sites are C1(x1,y1) and C2(x2,y2), and the weights of the two sites are α1 and α2 respectively, the coordinates of the merged cluster center are:

[0032]

[0033] Preferably, the fifth step uses a weighted site coordinate calculation method to merge sites that are too close together, and the site weight is determined by the number of services required by each site.

[0034] The beneficial effects of this invention are:

[0035] Compared with existing technologies, the multi-level station design method for customized buses with fixed stops but not fixed routes disclosed in this invention has the following advantages:

[0036] (1) The station design process and algorithm proposed in this invention can use the service parameters of the customized bus system as the control parameters of the station design, which enhances the operability and scientificity. The bus company can adjust the station design algorithm parameters according to passenger satisfaction feedback and operation status to improve service quality.

[0037] (2) The multi-level site clustering algorithm proposed in this invention is based on demand density clustering and can consider the limitation of site service radius throughout the entire algorithm process, completing the design of multi-level sites through multiple clustering. Specifically, the number of sites in a cluster is determined by buffer analysis, and the larger clusters generated in the first step of clustering are divided a second time, which can effectively avoid the problem in the prior art that it is easy to generate large clusters, resulting in sparse sites and long walking distances for passengers. Attached Figure Description

[0038] Figure 1 Flowchart illustrating the implementation of the multi-level station design method for fixed-station, non-fixed-route buses provided by this invention;

[0039] Figure 2 This is a distribution map of the starting point of demand in the data provided in the embodiments of the present invention;

[0040] Figure 3 This is a distribution map of demand endpoints in the data provided in this embodiment of the invention;

[0041] Figure 4 This is a diagram showing the calculation results of the main boarding station location provided in this embodiment of the invention;

[0042] Figure 5 This is a diagram showing the calculation results of the main drop-off station location provided in this embodiment of the invention;

[0043] Figure 6 Here is a schematic diagram of the buffer analysis principle in the site design method provided by this invention;

[0044] Figure 7 This is a diagram showing the calculation results of the boarding sub-station location provided in this embodiment of the invention;

[0045] Figure 8 This is a diagram showing the calculation results of the alighting sub-station location provided in an embodiment of the present invention;

[0046] Figure 9 This is a diagram showing the calculation results of the main and sub-site road network locations provided in an embodiment of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] See Figure 1 This invention discloses a multi-level station design method for customized buses with fixed stops but not fixed routes, specifically including the following steps:

[0049] The first step is to collect potential demand data. In this embodiment, ride-hailing booking data during the morning rush hour on a certain weekday is used as the potential demand data sample set; city vector maps and road network data are collected as map datasets.

[0050] The second step is to extract travel demand information from the order, including the origin and destination coordinates, departure time, and spatial coordinate system WGS 1984. The origin and destination distributions are as follows: Figure 2 and Figure 3 As shown.

[0051] Step 3: Main Site Design

[0052] S1: Spatial Clustering of Travel Demand Origin and Destination: Based on the latitude and longitude coordinates of the origin and destination of the demand, a density-based noise spatial clustering algorithm (DBSCAN) is adopted. Using the design station service radius R and the minimum cluster demand quantity minPts as parameters, spatial clustering is performed to generate origin and destination clusters of the demand. The service radius R is set to 300 meters based on the passenger walking distance, and the minimum demand quantity minPt is set based on the demand quantity corresponding to the minimum occupancy rate. In this embodiment, an 18-seat bus model is selected, and the minimum occupancy rate is set to 50%. Therefore, the parameter minPts is set to 18 × 50% = 9.

[0053] S2: Calculate the coordinates of the main stations: Using the K-means clustering algorithm, the required start and end points in each cluster are used as samples. The number of clusters, K, is set to 1. The centroid coordinates of each cluster are the coordinates of the main stations. The spatial distribution of the boarding and alighting main stations is as follows: Figure 4 and Figure 5 As shown, a total of 32 boarding master stations and 31 alighting master stations were generated.

[0054] Step 4, Subsite Division:

[0055] S1: Calculate the service area of ​​each main site based on buffer analysis: First, project each site, setting the projection coordinate system to WGS 1984 Web Mercator Auxiliary; generate circular buffers for the starting and ending points of demand in each main site cluster, with a radius set to the site service radius R = 300 meters, and merge the demand buffers in the same cluster (see the schematic diagram of the fusion principle). Figure 6 As shown in the figure, the area of ​​the merged buffer zone is the service area of ​​each boarding and alighting station.

[0056] S2: Calculate the number of sub-sites: Calculate the area A of the service area of ​​each main site, and determine the number Q of sub-sites that each main site demand cluster needs to be divided into based on the following formula:

[0057]

[0058] In this embodiment, the minimum service radius R is selected as 300 meters; π is the mathematical constant pi, which is taken as 3.14 in this embodiment.

[0059] S3: Sub-site Demand Re-clustering: Using the K-means clustering algorithm, each demand point in its respective cluster is used as a sample. The number of clusters is set to K = Q, resulting in Q sub-demand clusters for each main site. The coordinates of the cluster center for each sub-demand cluster are then calculated. In this embodiment, the calculated cluster center locations for the sub-demand clusters are as follows: Figure 7 and 8 As shown, where, Figure 7The central position of the demand cluster is the starting point of the demand. Figure 8 The center position of the demand destination sub-demand cluster is the location of the generated demand cluster. The center position of each generated demand cluster is the location of the alternative boarding and alighting stations, with a total of 153 boarding sub-stations and 154 alighting sub-stations generated.

[0060] Step 5: Deduplicating sites:

[0061] S1: Analyze the spatial distribution of the boarding and alighting sub-stations generated at the starting and ending points of the demand, calculate the distance between each boarding sub-station and each alighting sub-station, and mark the stations with a station spacing less than the maximum service radius R = 300 meters as stations to be adjusted;

[0062] S2: Merge the sites to be adjusted using a center coordinate calculation method based on demand weights.

[0063] Assuming the coordinates of the two cluster center points are C1(x1,y1) and C2(x2,y2), and the weights of the two cluster center points are α1 and α2 respectively, the coordinates of the merged cluster center points are:

[0064]

[0065] The cluster center weights are determined based on the number of service requests provided by each site. For example, if site C1 serves 10 boarding requests, its weight α1 = 10; if site C2 serves 7 alighting requests, its weight α2 = 7. In this embodiment, there are 76 pairs of boarding and alighting sub-sites that are less than the minimum service radius R, resulting in a total of 231 sub-sites after merging.

[0066] Step 6: Calculate the location of the sub-stations: Project the deduplicated sub-stations onto the road network to obtain the final geographical locations of the customized bus pick-up and drop-off points.

[0067] Each deduplicated sub-site is projected onto the nearest road segment and road network node, where the road network includes arterial roads, secondary arterial roads, and local roads. The resulting locations of the main and sub-sites are as follows: Figure 9 As shown, the pentagram represents the location of the main site, and the pins represent the locations of each sub-site.

Claims

1. A method for designing multi-level bus stops for customized buses with fixed stops but variable routes, characterized in that: The steps are as follows: Step 1: Data Collection: Collect potential demand data, including ride-hailing platform order data and public transport IC card data. Fields that must be included or estimated in the travel big data include the origin location, destination location, and departure time timestamp; extract vector geographic data of the target area as a map dataset; Step 2: Extracting travel demand information: Extracting or calculating travel or order demand information, including the origin and destination coordinates and departure time; Step 3: Main Site Division: S1: Spatial clustering of travel demand origin and destination: Based on the origin and destination coordinates of the demand, the DBSCAN clustering algorithm is adopted, and spatial clustering is performed with the service radius R of the design station and the number of demands minPts corresponding to the minimum occupancy rate as parameters to generate the origin cluster and destination cluster of the demand. S2: Calculate the spatial coordinates of the main site: Using the K-means clustering algorithm, the starting and ending points of the demand in each cluster are used as samples. The number of clusters K is set to 1. Clustering is used to obtain the demand clusters. The centroid coordinates of each cluster are the coordinates of the main site. Step 4: Subsite division: S1: Calculate the service area of ​​each main site based on buffer analysis: Generate circular buffers as demand buffers for the demand start and end points in the demand cluster, with the radius set as the site service radius R, and merge the demand buffers in the same cluster to obtain the service area of ​​each main site; S2: Calculate the number of sub-sites: Calculate the service area A of each main site, and calculate the number of sub-sites Q to be divided based on the following formula: S3: Sub-site demand re-clustering: The K-means clustering algorithm is used with the start and end points of demand in each cluster as samples, and the number of clusters is set to Q. The sub-demand clusters of each main site are obtained by clustering, and the centroid of the sub-site demand clusters is obtained. Step 5: Deduplication of sites: S1: Analyze the spatial distribution of the boarding and alighting sub-stations generated at the starting and ending points of the demand; calculate the distance between each boarding sub-station and each alighting sub-station, and mark the stations with a station spacing less than the maximum service radius R as stations to be adjusted; S2: Merge the sites to be adjusted using a center coordinate calculation method based on demand weights. Step 6: Calculate the location of the sub-site: Project the cluster center of the deduplicated sub-site demand cluster onto the road network to obtain the actual site location coordinates.

2. The method for designing multi-level stops for customized buses with fixed stops but variable routes according to claim 1, characterized in that, The fifth step uses a weighted site coordinate calculation method to merge sites that are too close together. Assuming the coordinates of two sites are C1(x1,y1) and C2(x2,y2), and their weights are α1 and α2 respectively, the coordinates of the merged cluster centers are: Site weight is determined by the demand for services provided by each site.

Images