A bus network planning method based on taxi GPS data mining
Through the method based on taxi GPS data, the problem of inaccurate and inability to adapt to rapid changes in traditional bus network planning information is solved, and more accurate and adaptable bus network planning is achieved. By dividing grid units and establishing passenger flow and time matrix , and use biological heuristic algorithms to form a bus network.
Patent Information
- Application Number
- CN202111275636.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-10-29
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional bus network planning relies on human surveys, resulting in inaccurate information and inability to adapt to rapidly changing urban road networks, requiring a planning method based on real-world human mobile data.
Using a method based on taxi GPS data, we divide grid units, establish passenger flow and time matrix, and use biological heuristic algorithms to form a bus network, including ant colony algorithm to optimize routes.
Reduce costs, the accuracy and comprehensiveness of obtaining information can better reflect real-world human mobile data, and generate more adaptable bus network planning.
Smart Images

Figure CN113919586B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of public transportation network planning, and in particular to a public transportation network planning method based on taxi GPS data mining. Background Art
[0002] Traditional bus network planning and design is usually based on manual surveys, which takes a lot of time and human resources. In addition, the traditional method relies heavily on the experience of designers, so the information obtained from manual surveys is sometimes not accurate enough.
[0003] Today, the road networks of rapidly developing cities are constantly being constructed and optimized. Survey-based planning cannot adapt to the rapid changes in the road networks of large cities, but these changes will be reflected in the mobility patterns of urban residents. Therefore, there is an urgent need for a bus network planning method based on a large amount of real-world human mobility data. Summary of the invention
[0004] The present invention aims to at least solve the technical problems existing in the prior art, and in particular innovatively proposes a bus network planning method based on taxi GPS data mining.
[0005] In order to achieve the above-mentioned object of the present invention, the present invention provides a bus network planning method based on taxi GPS data mining, comprising the following steps:
[0006] S1, grid cell division: the study area is divided into grid cells of equal size, and the grid cells containing the taxi passengers’ boarding and alighting records are called hotspot grid cells;
[0007] S2, using the hotspot grid cells obtained in S1 and the PDR within the radius r to determine the candidate bus stops, where PDR represents the taxi passenger boarding and alighting records;
[0008] S3, establish passenger flow matrix and time matrix: according to the candidate bus stops and taxi GPS data obtained in S2, establish the passenger flow matrix FM and time matrix TM between the candidate bus stops;
[0009] S4, using biologically inspired algorithms to form a public transportation network.
[0010] In a preferred embodiment of the present invention, S2 comprises the following steps:
[0011] S2-1, calculate the pdr value of each hotspot grid cell;
[0012] S2-2, according to the descending order of pdr value, select the hotspot grid unit with the largest pdr value as a candidate site; put the candidate site into the candidate site set;
[0013] S2-3, taking the candidate site generated in step S2-2 as the center and r as the radius, delete the PDR within the range;
[0014] S2-4, update the pdr values of the remaining hotspot grid cells;
[0015] S2-5, determining whether there is still a hotspot grid unit whose pdr value is greater than the set pdr threshold, if so, jump to step S2-2; if not, execute the next step;
[0016] S2-6, obtain the final set of candidate sites;
[0017] Among them, PDR means taxi passenger boarding and alighting record;
[0018] r represents the service radius of a bus station;
[0019] pdr represents the number of PDRs within a hotspot grid cell within a radius of r.
[0020] In a preferred embodiment of the present invention, S3 includes:
[0021] In the passenger flow matrix FM, each element f ij represents the passenger flow from candidate station i to candidate station j;
[0022] In the time matrix TM, each element t ij represents the average time from candidate site i to candidate site j;
[0023] The value of each element in the matrix is determined according to the taxi GPS data.
[0024] In a preferred embodiment of the present invention, S4 comprises the following steps:
[0025] S4-1, taking the passenger flow matrix and time matrix as input, using the biological heuristic algorithm to obtain an optimal route R best ;
[0026] S4-2, update the passenger flow matrix: replace R in the passenger flow matrix FM best The passenger flow between each station is set to 0;
[0027] S4-3, determining whether the number of generated bus routes is sufficient, if not, continuing to execute step S4-1;
[0028] S4-4, all the generated routes form a bus network.
[0029] In a preferred embodiment of the present invention, the bio-inspired algorithm comprises the following steps:
[0030] SA, takes the passenger flow matrix and time matrix as input and initializes the parameters of the ant colony algorithm;
[0031] SB, if the number of iterations ≤ iter max , execute the next step, otherwise end; where iter max Indicates the number of iterations of the ant colony algorithm;
[0032] SC, each ant generates a route: for the lth ant, the next site is randomly selected according to the transfer probability and added to the current route; the total time T of the current route is calculated, if T < T max , the current ant continues to search for the next site, otherwise, the current ant stops searching, saves the current route, and switches to the next ant to search for the route until m ants generate m routes; where T max is the bus route time constraint, m represents the total number of ants;
[0033] SD, for the m generated routes, calculate the total passenger flow of each route and update the optimal route;
[0034] SE, update the pheromone between candidate sites, increase the number of iterations by 1, and execute step SB.
[0035] In a preferred embodiment of the present invention, the formula for the transition probability in the SC is:
[0036]
[0037] in, represents the probability that the lth ant moves from candidate site i to candidate site j;
[0038] τ ij represents the pheromone from candidate site i to candidate site j;
[0039] τ ik represents the pheromone from candidate site i to candidate site k;
[0040] η ij represents the passenger flow between candidate site i and candidate site j;
[0041] η ik represents the passenger flow between candidate site i and candidate site k;
[0042] α represents the pheromone weight value;
[0043] β represents the visibility weight value;
[0044] Allowed represents the set of optional next stops.
[0045] In a preferred embodiment of the present invention, the formula for calculating the total time of the current route in the SC is:
[0046]
[0047] Where T represents the total time of the current route in both directions;
[0048] n represents the total number of stops on the route;
[0049] tm(s i ,s i+1 ) indicates that from s i to i+1 The average time
[0050] tm(s i+1 ,s i ) indicates that from s i+1 to i The average time
[0051] s i represents the i-th station on the route;
[0052] s i+1 represents the i+1th station on the route;
[0053] t0 represents the time the bus stops at each station.
[0054] In a preferred embodiment of the present invention, the formula for calculating the total passenger flow of each route in the SD is:
[0055]
[0056] Among them, Num represents the total passenger flow of the route in both directions;
[0057] n represents the total number of stops on the route;
[0058] fm(s i ,s j ) indicates that from s i to j passenger flow;
[0059] fm(s j ,s i ) indicates that from s j to i passenger flow;
[0060] s i represents the i-th station on the route;
[0061] s j represents the jth station on the route.
[0062] In a preferred embodiment of the present invention, the formula for updating pheromones between candidate sites in the SE is:
[0063] τ ij (t+1)=(1-ρ)·τ ij (t)+Δτ ij
[0064]
[0065]
[0066] Among them, τ ij (t+1) represents the pheromone from candidate site i to candidate site j at time t+1;
[0067] ρ represents the pheromone volatility coefficient;
[0068] τ ij (t) represents the pheromone from candidate site i to candidate site j at time t;
[0069] Δτ ij The pheromone increment from candidate site i to candidate site j;
[0070] m represents the total number of ants;
[0071] It means the pheromone released by the lth ant;
[0072] Q is the pheromone constant;
[0073] Num represents the passenger flow of the route in both directions.
[0074] In summary, due to the adoption of the above-mentioned technical scheme, compared with the bus network planning generated based on human survey information, the present invention can obtain human mobility data in the real world at a lower cost and with more comprehensive and accurate information, and generate a bus network planning based on this.
[0075] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0076] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:
[0077] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION
[0078] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.
[0079] The present invention proposes a bus network planning method based on taxi GPS data mining, comprising the following steps:
[0080] S1, grid unit division: the study area is divided into grid units of equal size, for example, 25m×25m grid units, and the grid units containing the taxi passengers’ boarding and alighting records are called hotspot grid units;
[0081] S2, using the hotspot grid cells obtained in S1 and the PDR within the radius r to determine the candidate bus stops, i.e., the passenger boarding and alighting hotspot areas, where PDR represents the taxi passenger boarding and alighting records;
[0082] S3, establish passenger flow matrix and time matrix: according to the candidate bus stops and taxi GPS data obtained in S2, establish the passenger flow matrix FM and time matrix TM between the candidate bus stops;
[0083] Taxi GPS data includes the time when passengers get on and off the taxi, the longitude and latitude of the location, etc.
[0084] S4, using biologically inspired algorithms to form a public transportation network.
[0085] In this paper, candidate stations refer to candidate bus stations.
[0086] In a preferred embodiment of the present invention, S2 comprises the following steps:
[0087] S2-1, calculate the pdr value of each hotspot grid cell;
[0088] S2-2, according to the descending order of pdr value, select the hotspot grid unit with the largest pdr value as a candidate site; put the candidate site into the candidate site set;
[0089] S2-3, taking the candidate site generated in step S2-2 as the center and r as the radius, delete the PDR within the range;
[0090] S2-4, update the pdr values of the remaining hotspot grid cells;
[0091] S2-5, determining whether there is still a hotspot grid unit whose pdr value is greater than the set pdr threshold, if so, jump to step S2-2; if not, execute the next step;
[0092] S2-6, obtain the final set of candidate sites;
[0093] Among them, PDR represents the taxi passenger boarding and alighting record;
[0094] r represents the service radius of a bus stop, which is generally 300m to 500m;
[0095] pdr represents the number of PDRs within a hotspot grid cell within a radius of r.
[0096] In a preferred embodiment of the present invention, S3 includes:
[0097] In the passenger flow matrix FM, each element f ij Represents the passenger flow from candidate site i to candidate site j; if the starting point of a taxi trip record is within the radius r of candidate site i and the end point is within the radius r of candidate site j, then it represents the passenger flow from candidate site i to candidate site j.
[0098] In the time matrix TM, each element t ij represents the average time from candidate station i to candidate station j; considering the speed difference between taxis and buses, we multiply the average time of taxis from candidate station i to candidate station j by λ to represent the average time of buses from candidate station i to candidate station j. λ is the ratio of the average speed of taxis to the average speed of buses in the city, and λ can be taken as 1.5.
[0099] The value of each element in the matrix is determined according to the taxi GPS data.
[0100] In a preferred embodiment of the present invention, S4 comprises the following steps:
[0101] S4-1, taking the passenger flow matrix and time matrix as input, using the biological heuristic algorithm to obtain an optimal route R best =[s1,s2,...s n ].
[0102] S4-2, update the passenger flow matrix: replace R in the passenger flow matrix FM best The passenger flow between each station is set to 0, that is: in represents any, n represents the total number of stops on the route, f(s i ,s j ) indicates that from s i to j Passenger flow i represents the i-th station on the route, s j represents the jth station on the route.
[0103] S4-3, determining whether the number of generated bus routes is sufficient, if not, continuing to execute step S4-1;
[0104] S4-4, all the generated routes form a bus network.
[0105] In a preferred embodiment of the present invention, the bio-inspired algorithm comprises the following steps:
[0106] SA, takes the passenger flow matrix and time matrix as input and initializes the parameters of the ant colony algorithm;
[0107] SB, if the number of iterations ≤ iter max , execute the next step, otherwise end;
[0108] where iter max Indicates the number of iterations of the ant colony algorithm;
[0109] SC, each ant generates a route: for the lth ant, randomly select the next site according to the transfer probability, l∈{1,2,3…m}, m represents the total number of ants, and add the site to the current route; calculate the total time T of the current route, if T <T max , the current ant continues to search for the next site, otherwise, the current ant stops searching, saves the current route, and switches to the next ant to search for the route until m ants generate m routes; where T max Time constraints for bus routes;
[0110] SD, for the m generated routes, calculate the total passenger flow of each route and update the optimal route;
[0111] SE, update the pheromone between candidate sites, increase the number of iterations by 1, and execute step SB.
[0112] In a preferred embodiment of the present invention, the formula of the transition probability in the SC is:
[0113]
[0114] in, represents the probability that the lth ant moves from candidate site i to candidate site j;
[0115] τ ij represents the pheromone from candidate site i to candidate site j;
[0116] τ ik represents the pheromone from candidate site i to candidate site k;
[0117] η ijrepresents the passenger flow between candidate site i and candidate site j;
[0118] η ik represents the passenger flow between candidate site i and candidate site k, that is, η ij =fm(i,j)+fm(j,i);
[0119] α represents the pheromone weight value;
[0120] β represents the visibility weight value;
[0121] Allowed represents the set of optional next stops.
[0122] In a preferred embodiment of the present invention, the formula for calculating the total time of the current route in the SC is:
[0123]
[0124] Where T represents the total time of the current route in both directions;
[0125] n represents the total number of stops on the route;
[0126] tm(s i ,s i+1 ) indicates that from s i to i+1 The average time
[0127] tm(s i+1 ,s i ) indicates that from s i+1 to i The average time
[0128] s i represents the i-th station on the route;
[0129] s i+1 represents the i+1th station on the route;
[0130] t0 represents the time the bus stops at each station.
[0131] In a preferred embodiment of the present invention, the formula for calculating the total passenger flow of each route in the SD is:
[0132]
[0133] Among them, Num represents the total passenger flow of the route in both directions;
[0134] n represents the total number of stops on the route;
[0135] fm(s i ,sj ) indicates that from s i to j passenger flow;
[0136] fm(s j ,s i ) indicates that from s j to i passenger flow;
[0137] s i represents the i-th station on the route;
[0138] s j represents the jth station on the route.
[0139] In a preferred embodiment of the present invention, the formula for updating pheromones between candidate sites in the SE is:
[0140] τ ij (t+1)=(1-ρ)·τ ij (t)+Δτ ij
[0141]
[0142]
[0143] Among them, τ ij (t+1) represents the pheromone from candidate site i to candidate site j at time t+1;
[0144] ρ represents the pheromone volatility coefficient;
[0145] τ ij (t) represents the pheromone from candidate site i to candidate site j at time t;
[0146] Δτ ij The pheromone increment from candidate site i to candidate site j;
[0147] m represents the total number of ants;
[0148] It means the pheromone released by the lth ant;
[0149] if the kth ant passes edge(i,j) means if the lth ant passes candidate site i to candidate site j;
[0150] Q is the pheromone constant;
[0151] Num represents the passenger flow of the route in both directions.
[0152] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.
Claims
1. A bus network planning method based on taxi GPS data mining, It is characterized in that The following steps are involved: S1, grid cell division: the study area is divided into grid cells of equal size, and the grid cells containing the taxi passengers’ boarding and alighting records are called hotspot grid cells; S2, using the hotspot grid cells obtained in S1 and the PDR within the radius r to determine the candidate bus stops, where PDR represents the taxi passenger boarding and alighting records; S2-1, calculate the pdr value of each hotspot grid cell; S2-2, according to the descending order of pdr value, select the hotspot grid unit with the largest pdr value as a candidate site; put the candidate site into the candidate site set; S2-3, taking the candidate site generated in step S2-2 as the center and r as the radius, delete the PDR within the range; S2-4, update the pdr values of the remaining hotspot grid cells; S2-5, determining whether there is still a hotspot grid unit whose pdr value is greater than the set pdr threshold, if so, jump to step S2-2; if not, execute the next step; S2-6, obtain the final set of candidate sites; Among them, PDR represents the taxi passenger boarding and alighting record; r represents the service radius of a bus station; pdr represents the number of PDRs within a radius of r in a hotspot grid cell; S3, establish the passenger flow matrix and time matrix: According to the candidate bus stops and taxi GPS data obtained in S2, establish the passenger flow matrix FM and time matrix TM between the candidate bus stops: In the passenger flow matrix FM, each element f ij represents the passenger flow from candidate station i to candidate station j; In the time matrix TM, each element t ij represents the average time from candidate site i to candidate site j; The value of each element in the matrix is determined by the taxi GPS data; S4, using biologically inspired algorithms to form a public transportation network; S4-1, taking the passenger flow matrix and time matrix as input, using the biological heuristic algorithm to obtain an optimal route R best ; S4-2, update the passenger flow matrix: replace R in the passenger flow matrix FM best The passenger flow between each station is set to 0; S4-3, determining whether the number of generated bus routes is sufficient, if not, continuing to execute step S4-1; S4-4, all the generated routes form a bus network.
2. A bus network planning method based on taxi GPS data mining according to claim 1, It is characterized in that The bio-inspired algorithm comprises the following steps: SA, takes the passenger flow matrix and time matrix as input and initializes the parameters of the ant colony algorithm; SB, if the number of iterations ≤ iter max , execute the next step, otherwise end; where iter max Indicates the number of iterations of the ant colony algorithm; SC, each ant generates a route: for the lth ant, the next site is randomly selected according to the transfer probability and added to the current route; the total time T of the current route is calculated, if T < T max , the current ant continues to search for the next site, otherwise, the current ant stops searching, saves the current route, and switches to the next ant to search for the route until m ants generate m routes; where T max is the bus route time constraint, m represents the total number of ants; SD, for the m generated routes, calculate the total passenger flow of each route and update the optimal route; SE, update the pheromone between candidate sites, increase the number of iterations by 1, and execute step SB.
3. A bus network planning method based on taxi GPS data mining according to claim 2, It is characterized in that The formula for the transition probability in the SC is: in, represents the probability that the lth ant moves from candidate site i to candidate site j; τ ij represents the pheromone from candidate site i to candidate site j; τ ik represents the pheromone from candidate site i to candidate site k; η ij represents the passenger flow between candidate site i and candidate site j; η ik represents the passenger flow between candidate site i and candidate site k; α represents the pheromone weight value; β represents the visibility weight value; Allowed represents the set of optional next stops.
4. A bus network planning method based on taxi GPS data mining according to claim 2, It is characterized in that The formula for calculating the total time of the current route in the SC is: Where T represents the total time of the current route in both directions; n represents the total number of stops on the route; tm(s i ,s i+1 ) indicates that from s i to i+1 The average time tm(s i+1 ,s i ) indicates that from s i+1 to i The average time s i represents the i-th station on the route; s i+1 represents the i+1th station on the route; t 0 Indicates the time the bus stops at each station.
5. A bus network planning method based on taxi GPS data mining according to claim 2, It is characterized in that The formula for calculating the total passenger flow for each route in the SD is: Among them, Num represents the total passenger flow of the route in both directions; n represents the total number of stops on the route; fm(si,sj) means from si to s j passenger flow; fm(sj,si) represents the j Passenger flow to si; s i represents the i-th station on the route; s j represents the jth station on the route.
6. A bus network planning method based on taxi GPS data mining according to claim 2, It is characterized in that The formula for updating pheromones between candidate sites in the SE is: t ij (t+1)=(1-ρ)·τ ij (t)+Δτ ij Among them, τ ij (t+1) represents the pheromone from candidate site i to candidate site j at time t+1; ρ represents the pheromone volatility coefficient; τ ij (t) represents the pheromone from candidate site i to candidate site j at time t; Δτ ij The pheromone increment from candidate site i to candidate site j; m represents the total number of ants; It means the pheromone released by the first ant; Q is the pheromone constant; Num represents the passenger flow of the route in both directions.
