Static rebalancing scheduling method for shared electric scooters based on street network data

Abstract

The present invention relates to the technical field of static rebalancing scheduling of shared electric scooters, and discloses a static rebalancing scheduling method for shared electric scooters based on street network data. The method includes obtaining shared electric scooter travel data, scheduling area boundary data, and street network data; extracting the vehicle status of all shared electric scooters at the beginning and end; identifying virtual site coordinates, and calculating the vehicle demand for each virtual site; calculating the shortest distance between virtual sites; using the vehicle demand for each virtual site and the shortest distance between virtual sites as input, minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process as the optimization goal, and using a heuristic algorithm to solve the static rebalancing scheduling result of the shared electric scooter. The present invention optimizes the rebalancing scheduling of electric scooters based on the shared electric scooter travel big data, which can effectively reduce operating costs.

Description

Technical Field

[0001] The present invention relates to the technical field of static rebalancing scheduling of shared electric scooters, and in particular to a static rebalancing scheduling method for shared electric scooters based on street network data. Background Art

[0002] As an emerging micro-shared transportation mode, shared electric scooters have been used in daily real life. They can provide a more convenient way of travel, especially for short-distance travel or last-mile travel. They are usually small in size and flexible in operation, and can travel quickly in the city to avoid traffic congestion. The rebalancing scheduling of shared electric scooters helps to reasonably allocate shared electric scooter resources and better meet people's daily travel needs.

[0003] Shared electric scooters are similar to shared bicycles in that they can generally be parked and used anywhere, and are therefore dockless. Virtual stations should be considered when discussing shared electric scooters. In addition, general studies generally express the rebalancing distance between stations by calculating the Euclidean distance between different stations, rather than calculating it based on actual street data. In addition, the static rebalancing problem is generally divided into a single-vehicle problem and a multi-vehicle problem. The single-vehicle problem is considered to be a "single-item pickup and delivery traveling salesman problem", while the multi-vehicle problem is called a "single-item pickup and delivery vehicle routing problem."

[0004] Generally speaking, for micro-shared transportation, whether it is a single vehicle or a multi-vehicle problem, this is a large-scale problem. The use of precise algorithms is not only time-consuming, but also has limited search capabilities and is often unsolvable. With the development of big data and data mining technology, the travel data generated by the shared electric scooter system provides new ideas for further optimizing scooter operations, but how to perform static rebalancing scheduling of shared electric scooters is still a very challenging problem. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a static rebalancing scheduling method for shared electric scooters based on street network data.

[0006] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0007] A static rebalancing scheduling method for shared electric scooters based on street network data comprises the following steps:

[0008] Obtain shared electric scooter trip data, dispatch area boundary data, and street network data;

[0009] Extract the vehicle status of all shared electric scooters at the beginning and end based on the shared electric scooter trip data and dispatch area boundary data;

[0010] identifying virtual station coordinates according to the vehicle states of all shared electric scooters at the beginning and end, and calculating the vehicle demand of each virtual station;

[0011] Calculate the shortest distance between virtual stations based on virtual station coordinates and street network data;

[0012] Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicles or the time of the entire rebalancing process as the optimization goal, a heuristic algorithm is used to solve the static rebalancing scheduling result of shared electric scooters.

[0013] Preferably, extracting the vehicle states of all shared electric scooters at the beginning and end according to the shared electric scooter travel data and the dispatch area boundary data includes the following steps:

[0014] Merge shared electric scooter trip data and dispatch area boundary data to determine the departure and destination data of all shared electric scooters;

[0015] Based on the origin and destination data of all shared electric scooters, extract the locations of all shared electric scooters at the start and end.

[0016] Preferably, before fusing the shared electric scooter travel data and the dispatch area boundary data, the method further includes:

[0017] Remove outliers from shared electric scooter travel data and dispatch area boundary data.

[0018] Preferably, identifying virtual station coordinates according to the vehicle states of all shared electric scooters at the beginning and the end, and calculating the vehicle demand of each virtual station, comprises the following steps:

[0019] The maximum silhouette coefficient is calculated according to the positions of all shared electric scooters at the beginning and the end, and the number of clusters is determined according to the maximum silhouette coefficient;

[0020] According to the positions of all shared electric scooters at the beginning, the K-means clustering algorithm is used to identify the coordinates of virtual stations, and the number of electric scooters at each virtual station at the beginning is counted;

[0021] Match the locations of all shared electric scooters at the end with the virtual station coordinates, and count the number of electric scooters at each virtual station at the end;

[0022] The vehicle demand at each virtual station is calculated based on the number of e-scooters at each virtual station at the beginning and end.

[0023] Preferably, calculating the shortest distance between virtual sites according to the virtual site coordinates and street network data comprises the following steps:

[0024] Match the virtual site coordinates with the street network data to obtain the projection point of the virtual site on the street network;

[0025] Screening the projection points with the shortest projection distance of each virtual site on the street network, and interpolating the screened projection points into the street network;

[0026] The Dijkstra algorithm is used to solve the shortest distance between virtual sites in the street network.

[0027] Preferably, the static rebalancing scheduling result of the shared electric scooter includes:

[0028] Optimal routes, total route length, and total rebalancing time for static rebalancing scheduling of shared electric scooters.

[0029] Preferably, the vehicle demand of each virtual site and the shortest distance between the virtual sites are used as input, and minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process is used as the optimization goal. A heuristic algorithm is used to solve the static rebalancing scheduling result of the shared electric scooter, including the following steps:

[0030] Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process as the optimization goal, a TSP-based rebalancing algorithm is used to solve the scheduling result of a single rebalancing shared electric scooter.

[0031] Preferably, the vehicle demand of each virtual station and the shortest distance between the virtual stations are used as input, and minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process is used as the optimization goal. The TSP-based rebalancing algorithm is used to solve the scheduling result of a single rebalancing shared electric scooter, including the following steps:

[0032] Initialize the total length of the route for static rebalancing scheduling of shared electric scooters;

[0033] Set the parking lot as the starting point of the TSP problem and solve it to obtain a current route that visits all virtual stations and returns to the parking lot;

[0034] Calculate the vehicle demand of each virtual station under current conditions according to the current route and generate the current unbalanced demand set;

[0035] identifying a set of virtual sites with unbalanced demands based on a current set of unbalanced demands;

[0036] When the unbalanced demand set is not all zero, the current virtual site is selected from the set of virtual sites with unbalanced demands;

[0037] Determine a distance list between the current virtual site and other virtual sites based on the shortest distance between the virtual sites;

[0038] Other virtual stations in the distance list with negative vehicle demand or visited are marked as having infinite distance;

[0039] Filter the virtual site closest to the current virtual site in the marked distance list;

[0040] In the current route, the positions of the current virtual site and the selected virtual site are exchanged to generate a new route;

[0041] Calculate the vehicle demand of each virtual station under the current conditions according to the new route and generate a new unbalanced demand set;

[0042] Determine whether the number of non-zero values ​​in the new unbalanced demand set is greater than the number of non-zero values ​​in the current unbalanced demand set; if so, replace the current unbalanced demand set with the new unbalanced demand set, and update the virtual station set with unbalanced demand; otherwise, keep the current route unchanged;

[0043] When the unbalanced demand set is all zero, add a parking lot at the end of the current route to form the optimal route;

[0044] Traverse each continuous virtual site pair in the optimal route, accumulate the distances between the continuous virtual site pairs, and obtain the total length of the route;

[0045] The total rebalancing time is calculated based on the total length of the route and the average speed of the shared electric scooters.

[0046] Preferably, the vehicle demand of each virtual site and the shortest distance between the virtual sites are used as input, and minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process is used as the optimization goal. A heuristic algorithm is used to solve the static rebalancing scheduling result of the shared electric scooter, including the following steps:

[0047] Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicles or the time of the entire rebalancing process as the optimization goal, a multi-vehicle rebalancing greedy algorithm is used to solve the scheduling results of the scheduling plan for all shared electric scooters in the target area.

[0048] Preferably, the vehicle demand of each virtual station and the shortest distance between the virtual stations are used as input, and the total distance traveled by the rebalancing vehicles or the time of the entire rebalancing process is minimized as the optimization goal. The multi-vehicle rebalancing greedy algorithm is used to solve the scheduling plan of all shared electric scooters in the target area, including the following steps:

[0049] Set the parking lot location;

[0050] Initialize the list of unvisited virtual sites;

[0051] Build a current route starting from the parking lot and initialize the vehicle load of the virtual station;

[0052] When the unvisited virtual site list is not empty, the last virtual site of the current path is recorded;

[0053] When the vehicle load at the virtual station is zero, the current vehicle is returned to the parking lot, the route length of the current vehicle and the number of available vehicles are updated, and a new route is started;

[0054] The available virtual station list is filtered according to the unvisited virtual station list and the load of the current vehicle; when the available virtual station list is empty and the number of available vehicles is zero, the process ends; when the available virtual station list is empty and the number of available vehicles is not zero, the current vehicle is returned to the parking lot, the route length of the current vehicle and the number of available vehicles are updated, and a new route is started;

[0055] Sort each virtual site in the filtered available virtual site list by the distance from the current site list; traverse the sorted available virtual site list to determine the vehicle load of each virtual site; when the vehicle load of a virtual site is less than the vehicle capacity threshold, add the virtual site to the current route, update the route length of the current vehicle and the vehicle load of the virtual site, and remove the virtual site from the list of unvisited virtual sites;

[0056] When the list of unvisited virtual sites is empty, all paths are merged into the optimal path;

[0057] Add up the lengths of all paths to get the total length of the route;

[0058] The total rebalancing time is calculated based on the length of the longest single path and the average speed of the shared electric scooters.

[0059] The present invention has the following beneficial effects:

[0060] 1. The data-driven static rebalancing method for shared electric scooters designed by the present invention can use the huge amount of data collected when the shared electric scooter travel data is collected to optimize the rebalancing process of the electric scooter, improve the search efficiency and accuracy of the static rebalancing of the shared electric scooter, and relatively reduce the operating cost.

[0061] 2. The data-driven static rebalancing method for shared electric scooters designed by the present invention does not directly use the Euclidean distance when calculating the distance between each virtual station, but instead uses a street network search and matching method to obtain a distance closer to the actual distance, making the solution to the problem closer to reality and more instructive for real problems.

[0062] 3. The data-driven static rebalancing method for shared electric scooters designed in the present invention can be implemented in any urban scenario and for single or multiple rebalancing vehicles without changing the method. It is a general framework for rebalancing optimization of shared electric scooters, and has broad application prospects and scope. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A flowchart of a static rebalancing scheduling method for shared electric scooters based on street network data;

[0064] Figure 2 Schematic diagram of projection and interpolation of virtual sites. DETAILED DESCRIPTION

[0065] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0066] The present invention aims at the problem that the actual distance between stations cannot be solved by using Euclidean distance and the static rebalancing cannot be solved by precise algorithms. The present invention provides a static rebalancing scheduling method for shared electric scooters with single or multiple rebalancing vehicles based on actual street network data. This method is not only practical, but also integrates big data, data mining technology, and intelligent algorithms, and can provide new technical support for the operation of shared electric scooters, an emerging micro-shared transportation mode. Therefore, the innovation of the present invention lies in the use of data mining technology and intelligent optimization algorithms, and the use of existing data resources to propose a new static rebalancing method for shared electric scooters that is closer to reality and has an optional rebalancing vehicle scale.

[0067] like Figure 1 As shown, an embodiment of the present invention provides a static rebalancing scheduling method for shared electric scooters based on street network data, comprising the following steps S1 to S5:

[0068] S1, obtaining shared electric scooter trip data, dispatch area boundary data, and street network data;

[0069] S2, extracting the vehicle status of all shared electric scooters at the beginning and end according to the shared electric scooter travel data and the dispatch area boundary data;

[0070] In an optional embodiment of the present invention, the present embodiment extracts the vehicle states of all shared electric scooters at the beginning and end according to the shared electric scooter travel data and the dispatch area boundary data, including the following steps:

[0071] Merge shared electric scooter trip data and dispatch area boundary data to determine the departure and destination data of all shared electric scooters;

[0072] Based on the origin and destination data of all shared electric scooters, extract the locations of all shared electric scooters at the start and end.

[0073] Before fusing the shared electric scooter travel data and the dispatch area boundary data, this embodiment further includes:

[0074] Remove outliers from shared electric scooter travel data and dispatch area boundary data.

[0075] S3, identifying the virtual station coordinates according to the vehicle status of all shared electric scooters at the beginning and the end, and calculating the vehicle demand of each virtual station;

[0076] In an optional embodiment of the present invention, the embodiment identifies virtual station coordinates according to the vehicle states of all shared electric scooters at the beginning and the end, and calculates the vehicle demand of each virtual station, including the following steps:

[0077] The maximum silhouette coefficient is calculated according to the positions of all shared electric scooters at the beginning and the end, and the number of clusters is determined according to the maximum silhouette coefficient;

[0078] According to the positions of all shared electric scooters at the beginning, the K-means clustering algorithm is used to identify the coordinates of virtual stations, and the number of electric scooters at each virtual station at the beginning is counted;

[0079] Match the locations of all shared electric scooters at the end with the virtual station coordinates, and count the number of electric scooters at each virtual station at the end;

[0080] The vehicle demand at each virtual station is calculated based on the number of e-scooters at each virtual station at the beginning and end.

[0081] Specifically, this embodiment first uses the K-means clustering algorithm to identify virtual sites. In order to determine the optimal number of clusters, the maximum silhouette coefficient needs to be calculated first. The silhouette coefficient of each object i is calculated as follows:

[0082]

[0083] Among them, a(i) is the average distance from the i-th object to other objects in the same cluster; b(i) is the minimum average distance from the i-th object to objects in different clusters, which is minimized in the cluster.

[0084] This embodiment takes the dispatch area boundary data and the scooter location data at the beginning and end of the day as input. The virtual station identification and matching process is as follows:

[0085] (1) Obtain the coordinates of all electric scooters at the starting point of the study traffic area and try to calculate different silhouette coefficients.

[0086] (2) Select the value k with the largest silhouette coefficient as the optimal number of clusters.

[0087] (3) Cluster the coordinates of all electric scooters at the beginning to obtain virtual stations, and count the number of electric scooters at each station.

[0088] (4) Match the coordinates of all electric scooters at the end with the virtual stations and count the number of electric scooters at each station.

[0089] After obtaining the number of electric scooters at each virtual station at the beginning and the end, this embodiment can further calculate the demand of each station. The rebalancing demand of each station is obtained by subtracting the number of electric scooters at the beginning from the number of electric scooters at the end. If the demand of a station is negative, it means that there is a lack of electric scooters; if the demand is positive, it means that there are excess electric scooters. In addition, due to the existence of electric scooters traveling across zones, the number of electric scooters at the beginning and end of the traffic area may be different. To solve this problem, this embodiment sets up a parking lot equipped with enough electric scooters in each traffic zone.

[0090] S4, calculating the shortest distance between virtual sites according to the virtual site coordinates and street network data;

[0091] In an optional embodiment of the present invention, the shortest distance between virtual sites is calculated according to the virtual site coordinates and street network data, including the following steps:

[0092] Match the virtual site coordinates with the street network data to obtain the projection point of the virtual site on the street network;

[0093] Screening the projection points with the shortest projection distance of each virtual site on the street network, and interpolating the screened projection points into the street network;

[0094] The Dijkstra algorithm is used to solve the shortest distance between virtual sites in the street network.

[0095] Specifically, when calculating the actual distance between virtual sites, this embodiment does not directly use the Euclidean distance, but uses a street network search and matching method to obtain a distance closer to the actual distance. The specific steps are as follows:

[0096] (1) Match the virtual stations and yards with the actual street network. The coordinates of the virtual stations are the locations of each cluster center obtained by K-means clustering, while the coordinates of the yards are further obtained by clustering the coordinates of the virtual stations. Therefore, the locations of the virtual stations and yards are not necessarily all on the streets, and they need to be matched with the nearest streets. The specific matching method includes projecting the virtual stations and yards onto several nearby streets and finding the street distance with the shortest projection distance. The projected positions on the streets are then interpolated into the street network. Before projection and interpolation, the virtual stations, yards, and street networks are converted to the same 1984 World Geodetic System (WGS84) coordinate system.

[0097] (2) Calculate the shortest path. Use the Dijkstra algorithm to find the shortest path between sites in the interpolation network graph, and use the distance of the path as the shortest distance between each site.

[0098] Figure 2 The projection and interpolation process of virtual stations on the road network is shown. The black lines represent the actual streets, the circular points represent the nodes where these streets start, end or intersect with each other, the square points represent the virtual stations obtained by clustering, and the star points represent the square points interpolated to the streets.

[0099] S5. Taking the vehicle demand of each virtual site and the shortest distance between virtual sites as input, minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process as the optimization goal, a heuristic algorithm is used to solve the static rebalancing scheduling result of the shared electric scooter.

[0100] In an optional embodiment of the present invention, the static rebalancing scheduling result of the shared electric scooter in this embodiment includes:

[0101] Optimal routes, total route length, and total rebalancing time for static rebalancing scheduling of shared electric scooters.

[0102] This embodiment uses the calculated station demand and the distance between stations as input, and uses two heuristic algorithms to solve the rebalancing problem of shared electric scooters in the single-vehicle and multi-vehicle cases respectively. In the rebalancing of shared electric scooters, the following points need to be paid attention to: (1) All rebalancing vehicles depart from the parking lot and return to the parking lot. (2) If the total demand of each station in an area is not zero, it is necessary to pick up or deliver the vehicle from the parking lot. Create a pseudo parking lot node, the demand of the pseudo node corresponds to the number of electric vehicles picked up or delivered from the parking lot, and its position corresponds to the coordinates of the parking lot. (3) The goal of rebalancing is to minimize the total distance traveled by the rebalancing vehicles or the time of the entire rebalancing process. When calculating the rebalancing time, ignore the time required to pick up or drop off the electric scooter at the station.

[0103] In an optional embodiment of the present invention, the vehicle demand of each virtual site and the shortest distance between virtual sites are used as inputs, and the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process is minimized as the optimization goal. A heuristic algorithm is used to solve the static rebalancing scheduling result of the shared electric scooter, including the following steps:

[0104] Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process as the optimization goal, a TSP-based rebalancing algorithm is used to solve the scheduling result of a single rebalancing shared electric scooter.

[0105] This embodiment uses the vehicle demand of each virtual station and the shortest distance between virtual stations as input, takes minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process as the optimization goal, and adopts a TSP-based rebalancing algorithm to solve the scheduling result of a single rebalancing shared electric scooter, including the following steps:

[0106] Initialize the total length of the route for static rebalancing scheduling of shared electric scooters;

[0107] Set the parking lot as the starting point of the TSP problem and solve it to obtain a current route that visits all virtual stations and returns to the parking lot;

[0108] Calculate the vehicle demand of each virtual station under current conditions according to the current route and generate the current unbalanced demand set;

[0109] identifying a set of virtual sites with unbalanced demands based on a current set of unbalanced demands;

[0110] When the unbalanced demand set is not all zero, the current virtual site is selected from the set of virtual sites with unbalanced demands;

[0111] Determine a distance list between the current virtual site and other virtual sites based on the shortest distance between the virtual sites;

[0112] Other virtual stations in the distance list with negative vehicle demand or visited are marked as having infinite distance;

[0113] Filter the virtual site closest to the current virtual site in the marked distance list;

[0114] In the current route, the positions of the current virtual site and the selected virtual site are exchanged to generate a new route;

[0115] Calculate the vehicle demand of each virtual station under the current conditions according to the new route and generate a new unbalanced demand set;

[0116] Determine whether the number of non-zero values ​​in the new unbalanced demand set is greater than the number of non-zero values ​​in the current unbalanced demand set; if so, replace the current unbalanced demand set with the new unbalanced demand set, and update the virtual station set with unbalanced demand; otherwise, keep the current route unchanged;

[0117] When the unbalanced demand set is all zero, add a parking lot at the end of the current route to form the optimal route;

[0118] Traverse each continuous virtual site pair in the optimal route, accumulate the distances between the continuous virtual site pairs, and obtain the total length of the route;

[0119] The total rebalancing time is calculated based on the total length of the route and the average speed of the shared electric scooters.

[0120] Specifically, this embodiment adopts a single-vehicle rebalancing algorithm based on TSP (traveling salesman problem). First, the OR-Tools (operations research tool) software package is used to solve the cross-station TSP problem without considering the rebalancing requirements of each station. Then, the greedy algorithm is used to adjust and optimize the route. In the main loop of the algorithm, the total distance is minimized by continuously selecting the nearest station for exchange. The loop is repeated until the requirements of all stations are met, that is, the requirements of all stations are zero, and the optimal route, the total length of the route, and the total rebalancing time are returned. The specific process is as follows:

[0121] A1. Initialize the total length L of the route for static rebalancing of shared electric scooters best =0;

[0122] A2. Set the parking lot as the starting point of the TPS problem and solve it to obtain an initial route R that visits all virtual sites and returns to the parking lot. initial ;

[0123] A3. According to the initial route R initial Calculate the vehicle demand of each virtual station under current conditions and generate an initial unbalanced demand set It is expressed as:

[0124]

[0125] Among them, CheackDemand is the vehicle demand calculation function, D i is the demand list of each virtual station in the i-th traffic area;

[0126] A4. Based on the initial unbalanced demand set Identify virtual site collections with imbalanced demand

[0127] A5. Determine the initial unbalanced demand set Are all the values ​​in 0? If so, proceed to A14; otherwise, proceed to A6;

[0128] A6. Virtual site collection from unbalanced demand Select the first virtual site n first ;

[0129] A7. According to the distance matrix M between virtual sites i Determine the first virtual site n first List of distances to other virtual sites

[0130] A8. List the distances The vehicle demand is negative and other virtual stations are marked as having infinite distance;

[0131] A9. List the distances Other virtual sites that have been visited in are marked as having infinite distance;

[0132] A10, List of distances after marking Screening with the first virtual site n first The nearest virtual site n nearest ;

[0133] A11, on the initial route R initial Switch the first virtual site n first Virtual site with screening nearest The location generates a new route R swaped ;

[0134] A12, according to the new route R swaped Calculate the vehicle demand of each virtual station under current conditions and generate a new unbalanced demand set It is expressed as:

[0135]

[0136] A13. Determine the new imbalance demand set Is the number of non-zero values ​​in greater than the initial unbalanced demand set? The number of non-zero values ​​in; if so, the initial unbalanced demand set Replace with a new set of unbalanced requirements and update the virtual site collection with unbalanced demand Otherwise, keep the current route unchanged and return to A6;

[0137] A14, on the new route R swaped Add a parking lot at the end of to form the optimal route R best ;

[0138] A15. Traverse the optimal route R best For each consecutive virtual site pair in the , the distances between consecutive virtual site pairs are accumulated to obtain the total route length L best ;

[0139] A16. According to the total length of the route L best and the average speed v of the shared electric scooters to calculate the total rebalancing time T best , expressed as:

[0140] T best ←L best / v.

[0141] In an optional embodiment of the present invention, the vehicle demand of each virtual site and the shortest distance between virtual sites are used as inputs, and the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process is minimized as the optimization goal. A heuristic algorithm is used to solve the static rebalancing scheduling result of the shared electric scooter, including the following steps:

[0142] Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicles or the time of the entire rebalancing process as the optimization goal, a multi-vehicle rebalancing greedy algorithm is used to solve the scheduling plan for all shared electric scooters in the target area.

[0143] This embodiment uses the vehicle demand of each virtual station and the shortest distance between virtual stations as input, takes minimizing the total distance traveled by the rebalancing vehicles or the time of the entire rebalancing process as the optimization goal, and adopts a multi-vehicle rebalancing greedy algorithm to solve the scheduling plan for all shared electric scooters in the target area, including the following steps:

[0144] Set the parking lot location;

[0145] Initialize the list of unvisited virtual sites;

[0146] Build a current route starting from the parking lot and initialize the vehicle load of the virtual station;

[0147] When the unvisited virtual site list is not empty, the last virtual site of the current path is recorded;

[0148] When the vehicle load at the virtual station is zero, the current vehicle is returned to the parking lot, the route length of the current vehicle and the number of available vehicles are updated, and a new route is started;

[0149] The available virtual station list is filtered according to the unvisited virtual station list and the load of the current vehicle; when the available virtual station list is empty and the number of available vehicles is zero, the process ends; when the available virtual station list is empty and the number of available vehicles is not zero, the current vehicle is returned to the parking lot, the route length of the current vehicle and the number of available vehicles are updated, and a new route is started;

[0150] Sort each virtual site in the filtered available virtual site list by the distance from the current site list; traverse the sorted available virtual site list to determine the vehicle load of each virtual site; when the vehicle load of a virtual site is less than the vehicle capacity threshold, add the virtual site to the current route, update the route length of the current vehicle and the number of available vehicles, and remove the virtual site from the unvisited virtual site list;

[0151] When the list of unvisited virtual sites is empty, all paths are merged into the optimal path;

[0152] Add up the lengths of all paths to get the total length of the route;

[0153] The total rebalancing time is calculated based on the length of the longest single path and the average speed of the shared electric scooters.

[0154] Specifically, this embodiment adopts a multi-vehicle rebalancing greedy algorithm. First, a greedy strategy is used to construct a route, and the nearest stations are selected in turn to be added to the current route until all stations have been visited. When selecting the next station, the vehicle load and station demand conditions are considered to meet the vehicle capacity limit and station demand. When the rebalancing vehicle departs from the parking lot, after passing through several stations, when the vehicle load is zero, the vehicle returns to the parking lot and starts a new route again. During the route construction process, information such as route length and rebalancing time will be dynamically updated, and the complete route will be added to the solution. The generated solution is output, which contains multiple routes, each of which represents the driving route of a vehicle. In addition, the total rebalancing time is the rebalancing time of the longest route. The specific process is as follows:

[0155] B1. Set the parking lot location;

[0156] B2. Initialize the list of unvisited virtual sites

[0157] B3. Determine whether the parking lot demand is not 0; if so, set up a pseudo parking lot and add it to the list of unvisited virtual sites and set the parking lot demand to 0; otherwise proceed to B4;

[0158] B4, construct a current route r starting from the parking lot;

[0159] B5, initialize the vehicle load of the virtual site b = 0;

[0160] B6. Determine the list of virtual sites that have not been visited Is the length of not 0? If so, record the last virtual site n of the current path last ; Otherwise proceed to B14;

[0161] B7, determine whether the vehicle load of the virtual station is 0; if so, return the current vehicle to the parking lot, update the route length l of the current vehicle and the number of available vehicles k, and start a new route; otherwise, proceed to B8;

[0162] B8. Based on the list of unvisited virtual sites and the current vehicle load b filter the list of available virtual stations It is expressed as:

[0163]

[0164] in, To filter the list of unvisited virtual sites according to the current vehicle's load b These selected sites are likely to become the next site on the route.

[0165] B9. Determine the list of available virtual sites Is it not empty? If so, proceed to B10; otherwise, return the current vehicle to the parking lot, update the route length l of the current vehicle and the number of available vehicles k, and start a new route;

[0166] B10, determine whether the number of available vehicles k is 0; if so, end the process; otherwise, proceed to step B11;

[0167] B11. The list of available virtual sites to be screened The virtual sites in the list are sorted by their distance from the current site list to obtain a sorted virtual site list. It is expressed as:

[0168]

[0169] in, A list of virtual sites that will be available based on the distance matrix Each virtual site in the last virtual site n last A function to sort by distance;

[0170] B12. Traverse and sort the virtual site list Determine the vehicle load b at each virtual station tmp ;

[0171] B13. Determine the vehicle load of the virtual site b tmp Is it less than the vehicle's capacity threshold c? If so, add the virtual site to the current route r, update the route length l of the current vehicle and the vehicle load b of the virtual site, and add the virtual site to the list of unvisited virtual sites Removed;

[0172] B14. Merge all paths r into the optimal path R best ;

[0173] B15. Add up the lengths l of all paths to get the total length L of the route best ;

[0174] B16. Calculate the total rebalancing time T based on the length l of the longest single path and the average speed v of the shared electric scooters best .

[0175] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0176] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0177] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0178] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0179] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A static rebalancing scheduling method for shared electric scooters based on street network data, characterized in that: The following steps are involved: Obtain shared electric scooter trip data, dispatch area boundary data, and street network data; Extract the vehicle status of all shared electric scooters at the beginning and end based on the shared electric scooter trip data and dispatch area boundary data; including: Merge shared electric scooter trip data and dispatch area boundary data to determine the departure and destination data of all shared electric scooters; Based on the origin and destination data of all shared electric scooters, extract the locations of all shared electric scooters at the start and end; identifying virtual station coordinates according to the vehicle states of all shared electric scooters at the beginning and end, and calculating the vehicle demand of each virtual station; Calculate the shortest distance between virtual stations based on virtual station coordinates and street network data; Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process as the optimization goal, a heuristic algorithm is used to solve the static rebalancing scheduling results of shared electric scooters, including: Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicles or the time of the entire rebalancing process as the optimization goal, a TSP-based rebalancing algorithm is used to solve the scheduling plan for all shared electric scooters in the target area, including: Initialize the total length of the route for static rebalancing scheduling of shared electric scooters; Set the parking lot as the starting point of the TSP problem and solve it to obtain a current route that visits all virtual stations and returns to the parking lot; Calculate the vehicle demand of each virtual station under current conditions according to the current route and generate the current unbalanced demand set; identifying a set of virtual sites with unbalanced demands based on a current set of unbalanced demands; When the unbalanced demand set is not all zero, the current virtual site is selected from the set of virtual sites with unbalanced demands; Determine a distance list between the current virtual site and other virtual sites based on the shortest distance between the virtual sites; Other virtual stations in the distance list with negative vehicle demand or visited are marked as having infinite distance; Filter the virtual site closest to the current virtual site in the marked distance list; In the current route, the positions of the current virtual site and the selected virtual site are exchanged to generate a new route; Calculate the vehicle demand of each virtual station under the current conditions according to the new route and generate a new unbalanced demand set; Determine whether the number of non-zero values ​​in the new unbalanced demand set is greater than the number of non-zero values ​​in the current unbalanced demand set; if so, replace the current unbalanced demand set with the new unbalanced demand set, and update the virtual station set with unbalanced demand; otherwise, keep the current route unchanged; When the unbalanced demand set is all zero, add a parking lot at the end of the current route to form the optimal route; Traverse each continuous virtual site pair in the optimal route, accumulate the distances between the continuous virtual site pairs, and obtain the total length of the route; The total rebalancing time is calculated based on the total length of the route and the average speed of the shared electric scooters.

2. The static rebalancing scheduling method for shared electric scooters based on street network data according to claim 1 is characterized in that: Before fusing shared electric scooter trip data with dispatch area boundary data, it also includes: Remove outliers from shared electric scooter travel data and dispatch area boundary data.

3. The static rebalancing scheduling method for shared electric scooters based on street network data according to claim 1 is characterized in that: The virtual station coordinates are identified according to the vehicle states of all shared electric scooters at the beginning and the end, and the vehicle demand of each virtual station is calculated, including the following steps: The maximum silhouette coefficient is calculated according to the positions of all shared electric scooters at the beginning and the end, and the number of clusters is determined according to the maximum silhouette coefficient; According to the positions of all shared electric scooters at the beginning, the K-means clustering algorithm is used to identify the coordinates of virtual stations, and the number of electric scooters at each virtual station at the beginning is counted; Match the locations of all shared electric scooters at the end with the virtual station coordinates, and count the number of electric scooters at each virtual station at the end; The vehicle demand at each virtual station is calculated based on the number of e-scooters at each virtual station at the beginning and end.

4. The static rebalancing scheduling method for shared electric scooters based on street network data according to claim 1, characterized in that: Calculating the shortest distance between virtual stations based on virtual station coordinates and street network data includes the following steps: Match the virtual site coordinates with the street network data to obtain the projection point of the virtual site on the street network; Screening the projection points with the shortest projection distance of each virtual site on the street network, and interpolating the screened projection points into the street network; The Dijkstra algorithm is used to solve the shortest distance between virtual sites in the street network.

5. The static rebalancing scheduling method for shared electric scooters based on street network data according to claim 1, characterized in that: The static rebalancing scheduling results of shared electric scooters include: Optimal routes, total route length, and total rebalancing time for static rebalancing scheduling of shared electric scooters.

6. The static rebalancing scheduling method for shared electric scooters based on street network data according to claim 1, characterized in that: Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicle or the time of the entire rebalancing process as the optimization goal, a heuristic algorithm is used to solve the static rebalancing scheduling result of the shared electric scooter, including the following steps: Taking the vehicle demand of each virtual station and the shortest distance between virtual stations as input, minimizing the total distance traveled by the rebalancing vehicles or the time of the entire rebalancing process as the optimization goal, a multi-vehicle rebalancing greedy algorithm is used to solve the scheduling plan for all shared electric scooters in the target area, including: Set the parking lot location; Initialize the list of unvisited virtual sites; Build a current route starting from the parking lot and initialize the vehicle load of the virtual station; When the unvisited virtual site list is not empty, the last virtual site of the current path is recorded; When the vehicle load at the virtual station is zero, the current vehicle is returned to the parking lot, the route length of the current vehicle and the number of available vehicles are updated, and a new route is started; The list of available virtual sites is filtered according to the list of unvisited virtual sites and the load of the current vehicle; when the list of available virtual sites is empty and the number of available vehicles is zero, the process ends; when the list of available virtual sites is empty and the number of available vehicles is not zero, the current vehicle is returned to the parking lot, the route length of the current vehicle and the number of available vehicles are updated, and a new route is started; Sort each virtual site in the filtered available virtual site list by the distance from the current site list; traverse the sorted available virtual site list to determine the vehicle load of each virtual site; when the vehicle load of a virtual site is less than the vehicle capacity threshold, add the virtual site to the current route, update the route length of the current vehicle and the vehicle load of the virtual site, and remove the virtual site from the list of unvisited virtual sites; When the list of unvisited virtual sites is empty, all paths are merged into the optimal path; Add up the lengths of all paths to get the total length of the route; The total rebalancing time is calculated based on the length of the longest single path and the average speed of the shared electric scooters.

Images