A driver-passenger matching satisfaction method based on multi-stage bipartite graph
Through the multi-stage binary graph method, priority matches drivers with high altitude driving time and passengers with historical high waiting time, adjusts the price and commission ratio of remote orders, and solves the problem that drivers and passenger satisfaction in the existing technology has not been maximized, and realizes that More fair and efficient matching has been achieved, the user complaint rate has been reduced, and the platform has been promoted.
Patent Information
- Application Number
- CN202111193352.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-10-13
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2041-10-13
AI Technical Summary
The driver and passenger order allocation algorithm of the existing online shared travel platform failed to effectively maximize the satisfaction of drivers and passengers, resulting in long-term dissatisfaction among some people, frequent complaints, and unfair matching problems, affecting the development of the platform.
Using the multi-stage bipartite graph driver-passenger matching satisfaction method, establish a passenger and driver satisfaction model through optimization problems, set an optimization objective function, and prioritize the matching of drivers with high altitude driving time, passengers with historical high waiting time, and middle distance in stages Drivers and passengers, remote drivers and passengers adjust the order prices and driver commission ratios with order acceptance distances exceeding 3.5 kilometers to improve satisfaction and reduce cancellation rates.
It significantly improves the satisfaction of drivers and passengers, reduces the user complaint rate, balances the rights and interests of drivers and passengers, increases stickiness and trust in the platform, and promotes the development of online ride-hailing platforms.
Smart Images

Figure CN114065104B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of order matching between passengers and drivers in network shared travel, and in particular to a driver-passenger matching satisfaction method based on a multi-stage bipartite graph. Background Art
[0002] Currently, in the field of online shared travel, a key link is to assign the right driver to each passenger who takes a ride online. Different allocation algorithms will affect the driver's income, the number of orders on the online ride-hailing platform, and the passenger's experience. Drivers or passengers who have a bad experience for a long time will leave the online ride-hailing platform and seek other platforms. Based on this, a good driver allocation algorithm should take into account both passenger satisfaction and driver satisfaction, so as to increase the number of drivers who take orders and passengers who take rides, thereby promoting the increase in the number of orders on the online ride-hailing platform.
[0003] Currently, the industry mainly chooses to allocate orders to drivers and passengers globally based on GMV orientation or pick-up time orientation. The problem is that its method does not prioritize driver and passenger satisfaction, but optimizes a certain average indicator of the overall population.
[0004] The existing order allocation method optimized for average indicators is a typical unfair allocation algorithm, which will obey unfair allocation probability density functions such as normal distribution. According to probability theory, it will inevitably cause long-term dissatisfaction among some people, with passengers complaining about the difficulty of getting a taxi and drivers complaining about not receiving orders. Existing online car-hailing platforms have encountered such problems and cannot effectively solve them.
[0005] In actual situations, for most passengers, the satisfaction of a pick-up time of 3 minutes is similar to that of a pick-up time of 5 minutes. At this time, we should consider improving the satisfaction of other passengers. At the same time, for drivers who have not received orders for a long time, we should give them priority to increase their willingness to drive and avoid loss of transportation capacity. Therefore, the traditional method does not take driver and passenger satisfaction as the optimization goal, so it is difficult to maximize the satisfaction of drivers and passengers, which does not conform to the social development trend of a better life. Summary of the invention
[0006] The object of the present invention is to provide a driver-passenger matching satisfaction method based on a multi-stage bipartite graph to solve the problems raised in the above background technology.
[0007] To achieve the above object, the present invention provides the following technical solution: a driver-passenger matching satisfaction method based on a multi-stage bipartite graph, comprising the following specific steps:
[0008] S1. Optimization problem to establish a passenger satisfaction model;
[0009] S2, optimization problem to establish a satisfaction model for drivers;
[0010] S3, optimization problem setting optimization objective function;
[0011] S4. Initialize the bipartite graph G: Initially, there are n+m vertices, pi, dj, 1≤i≤n, 1≤j≤m. Initially, there are no edge connections between the vertices.
[0012] S5, in the first stage of the bipartite graph, drivers with high historical driving time and high pick-up time are prioritized for matching;
[0013] S6, the second stage of the bipartite graph, prioritizes matching passengers with historically high waiting times;
[0014] S7, the third stage of the bipartite graph, the mid-range driver and passenger satisfaction matching;
[0015] S8, the fourth stage of the bipartite graph, remote drivers and passengers’ satisfaction matching;
[0016] S9. Improve the satisfaction of passengers and drivers for this long-distance order: For the remote distance Sl set, check the matching orders whose order acceptance distance exceeds γ, γ = 3.5 kilometers, and adjust the passenger order price and the driver's commission ratio for these orders to improve satisfaction and reduce cancellation rate.
[0017] Preferably, step S1 is specifically as follows:
[0018] S1-1, let n passengers be assigned values from p1 to pn, and their geographical locations be recorded as locp1 to locpn; let m drivers be assigned values from d1 to dm, and their geographical locations be recorded as locd1 to locdm; f(pi,dj) is the estimated pick-up time of passenger pi and driver dj, in minutes, where 1≤i≤n, 1≤j≤m;
[0019] hp(pi) is the satisfaction of the passenger pi in this taxi ride, 1≤i≤n;
[0020] S1-2. Calculate hp(pi) according to the formula:
[0021] hp(pi)=(1-c(pi)) / (1+c(pi))×0.5+MAX((α-tp(pi)) / (α+tp(pi)),0)×0.5; where,
[0022] tp(pi) is the time it takes to pick up the passenger after the driver is finally assigned, in minutes, 1≤i≤n;
[0023] c(pi) is the percentage of the passenger's long-term taxi-hailing times in history, c(pi) = the number of times the passenger called for more than 5 minutes in 180 days / the total number of times the passenger took taxis in 180 days, 1≤i≤n;
[0024] α=e×today's average pick-up time. If no vehicle is assigned, tp(pi)=α, where e is a mathematical constant.
[0025] Preferably, step S2 is specifically as follows:
[0026] S2-1. Define hd(dj) as the satisfaction of the driver dj in this taxi ride, 1≤j≤m;
[0027] S2-2. Calculate hd(dj) according to the formula:
[0028] hd(dj)=(1-g(dj)) / (1+g(dj))×0.5+MAX((β-td(dj)) / (α+td(dj)),0)×0.5; among them,
[0029] g(dj) = (today's idle time + today's pick-up time) / number of orders received;
[0030] td(dj) is the pick-up time of the driver after the driver is finally assigned, in minutes, 1≤j≤m;
[0031] β = e × (today's average pick-up time + today's average idle time). If no passenger is assigned, td(dj) = β, where e is a mathematical constant.
[0032] Preferably, step S3 is specifically as follows:
[0033] The optimization function L = ∑hp(pi)+∑hd(dj), 1≤i≤n, 1≤j≤m. The problem is transformed into finding the matching set R and obtaining the maximum value of the function L; where (pi, dj) represents a matching pair of the R set, indicating the matching of pi and dj orders.
[0034] Preferably, step S5 is specifically as follows:
[0035] S5-1. Find a priority matching driver set, and give priority to matching drivers with high idle driving rate and high pick-up time. Specifically, sort the drivers in descending order according to g(dj), and select the drivers in the top 1 / 8 quartile to enter the priority matching driver set Sd;
[0036] S5-2. Construct the edge that preferentially matches the short neighbors of the driver. According to the set Sd, find the short neighbor set ShortNeighbourd(dj) for each driver. The neighbor is defined as the latitude and longitude locp and locd within 1 km. For the short neighbor set ShortNeighbourd(dj) of driver dj, construct the edge from pi to dj for graph G. The edge value is f(pi,dj), where pi∈ShortNeighbourd(dj).
[0037] S5-3. Solve the bipartite graph according to the Hungarian algorithm. The edge solved at this time is the match between the driver and the passenger. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
[0038] Preferably, step S6 is specifically as follows:
[0039] S6-1. Find a set of priority matching passengers, and give priority to matching passengers with a high historical waiting time. Specifically, sort the passengers in descending order according to c(pi), select the passengers in the first 1 / 8 quartile to enter the priority matching passenger set Sp. If the passenger has been matched in the first stage, remove it from the set;
[0040] S6-2. Construct the edge that preferentially matches the neighbors of passengers. According to the set Sp, find the short neighbor set Neighbourp(pi) for each passenger. The neighbors are defined as those within the range of 3 kilometers of longitude and latitude locp and locd. Here, we need to ensure that these passengers can get a taxi. The distance is set to 3 kilometers. For the neighbor set Neighbourp(pi) of passenger pi, if dj has been matched in the first stage, remove dj from Neighbourp(pi), and construct an edge from pi to dj for graph G. The edge value is f(pi,dj), where dj∈Neighbourp(pi);
[0041] S6-3. Solve the bipartite graph according to the Hungarian algorithm. The edge solved at this time is the match between the passenger and the driver. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
[0042] Preferably, step S7 is specifically as follows:
[0043] S7-1. Find the driver set USd that has not been matched yet;
[0044] S7-2. Construct the edge of the driver's short neighbors. According to the set USd, find the short neighbor set MiddleNeighbourd(dj) for each driver. The neighbor is defined as the latitude and longitude locp and locd within a range of 2 kilometers. For the neighbor set MiddleNeighbourd(dj) of driver dj, if pi has been matched in the previous stage, remove pi from MiddleNeighbourd(dj) and construct an edge from pi to dj for graph G. The edge value is f(pi,dj), where pi∈MiddleNeighbourd(dj);
[0045] S7-3. Solve the bipartite graph according to the Hungarian algorithm. The edge solved at this time is the match between the driver and the passenger. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
[0046] Preferably, step S8 is specifically as follows:
[0047] S8-1, find the driver set VSd that has not been matched yet;
[0048] S8-2. Construct the edge of the driver's long neighbors. According to the set VSd, find the short neighbor set Neighbourd(dj) for each driver. The neighbors are defined as those within the range of 5 kilometers of the longitude and latitude locp and locd. For the neighbor set Neighbourd(dj) of driver dj, if pi has been matched in the previous stage, remove pi from Neighbourd(dj) and construct an edge from pi to dj for graph G. The edge value is f(pi,dj), where pi∈Neighbourd(dj);
[0049] S8-3. Solve the bipartite graph according to the Hungarian algorithm. The edge solved at this time is the match between the driver and the passenger; remove other unmatched edges in the graph G. At this time, the edges in the graph G are the result set R, and only the matching set solved in the fourth stage is recorded as Sl.
[0050] Compared with the prior art, the present invention has the following beneficial effects:
[0051] 1. With the goal of maximizing driver and passenger satisfaction, through the multi-stage bipartite graph method, the first and second stages improve the historical satisfaction of passengers and drivers, and the third and fourth stages improve the current satisfaction of passengers and drivers. Compared with the existing technical methods, the value of the optimization function L is significantly improved, while reducing the user complaint rate.
[0052] 2. It solves the problem of drivers’ difficulty in accepting orders and users’ difficulty in hailing a taxi, optimizes the individual unfairness caused by probability problems in traditional algorithms, and balances the rights and interests of drivers and passengers. At the same time, the algorithm can improve the stickiness and trust of drivers and passengers in the travel platform, reduce reliance on luck, and form a virtuous cycle and word of mouth, thereby promoting the development of online car-hailing platforms.
[0053] 3. Compared with the traditional driver-passenger matching method, which will cause unfair matching effects due to probability and related complaints from drivers and passengers, the present invention has the advantages of automatically adjusting the satisfaction of drivers and passengers, fair matching, reducing user churn, etc., thereby enhancing the brand value and order volume of the online car-hailing platform. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] Figure 1 It is a schematic diagram of the overall process of the present invention. DETAILED DESCRIPTION
[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0056] See also Figure 1 The present invention provides a technical solution: a driver-passenger matching satisfaction method based on a multi-stage bipartite graph, comprising the following specific steps:
[0057] S1. Optimization problem to establish a passenger satisfaction model;
[0058] S2, optimization problem to establish a satisfaction model for drivers;
[0059] S3, optimization problem setting optimization objective function;
[0060] S4. Initialize the bipartite graph G: Initially, there are n+m vertices, pi, dj, 1≤i≤n, 1≤j≤m. Initially, there are no edge connections between the vertices.
[0061] S5, in the first stage of the bipartite graph, drivers with high historical driving time and high pick-up time are prioritized for matching;
[0062] S6, the second stage of the bipartite graph, prioritizes matching passengers with historically high waiting times;
[0063] S7, the third stage of the bipartite graph, the mid-range driver and passenger satisfaction matching;
[0064] S8, the fourth stage of the bipartite graph, remote drivers and passengers’ satisfaction matching;
[0065] S9. Improve the satisfaction of passengers and drivers for this long-distance order: For the remote distance Sl set, check the matching orders whose order acceptance distance exceeds γ, γ = 3.5 kilometers, and adjust the passenger order price and the driver's commission ratio for these orders to improve satisfaction and reduce cancellation rate.
[0066] Furthermore, step S1 is specifically as follows:
[0067] S1-1, let n passengers be assigned values from p1 to pn, and their geographical locations be recorded as locp1 to locpn; let m drivers be assigned values from d1 to dm, and their geographical locations be recorded as locd1 to locdm; f(pi,dj) is the estimated pick-up time of passenger pi and driver dj, in minutes, where 1≤i≤n, 1≤j≤m;
[0068] hp(pi) is the satisfaction of the passenger pi in this taxi ride, 1≤i≤n;
[0069] S1-2. Calculate hp(pi) according to the formula:
[0070] hp(pi)=(1-c(pi)) / (1+c(pi))×0.5+MAX((α-tp(pi)) / (α+tp(pi)),0)×0.5; where,
[0071] tp(pi) is the time it takes to pick up the passenger after the driver is finally assigned, in minutes, 1≤i≤n;
[0072] c(pi) is the percentage of the passenger's long-term taxi-hailing times in history, c(pi) = the number of times the passenger called for more than 5 minutes in 180 days / the total number of times the passenger took taxis in 180 days, 1≤i≤n;
[0073] α=e×today's average pick-up time. If no vehicle is assigned, tp(pi)=α, where e is a mathematical constant.
[0074] Further, step S2 is specifically as follows:
[0075] S2-1. Define hd(dj) as the satisfaction of the driver dj in this taxi ride, 1≤j≤m;
[0076] S2-2. Calculate hd(dj) according to the formula:
[0077] hd(dj)=(1-g(dj)) / (1+g(dj))×0.5+MAX((β-td(dj)) / (α+td(dj)),0)×0.5; among them,
[0078] g(dj) = (today's idle time + today's pick-up time) / number of orders received;
[0079] td(dj) is the pick-up time of the driver after the driver is finally assigned, in minutes, 1≤j≤m;
[0080] β = e × (today's average pick-up time + today's average idle time). If no passenger is assigned, td(dj) = β, where e is a mathematical constant.
[0081] Furthermore, step S3 is specifically as follows:
[0082] The optimization function L = ∑hp(pi)+∑hd(dj), 1≤i≤n, 1≤j≤m. The problem is transformed into finding the matching set R and obtaining the maximum value of the function L; where (pi, dj) represents a matching pair of the R set, indicating the matching of pi and dj orders.
[0083] Furthermore, step S5 is specifically as follows:
[0084] S5-1. Find a priority matching driver set, and give priority to matching drivers with high idle driving rate and high pick-up time. Specifically, sort the drivers in descending order according to g(dj), and select the drivers in the top 1 / 8 quartile to enter the priority matching driver set Sd;
[0085] S5-2. Construct the edge that preferentially matches the short neighbors of the driver. According to the set Sd, find the short neighbor set ShortNeighbourd(dj) for each driver. The neighbor is defined as the latitude and longitude locp and locd within 1 km. For the short neighbor set ShortNeighbourd(dj) of driver dj, construct the edge from pi to dj for graph G. The edge value is f(pi,dj), where pi∈ShortNeighbourd(dj).
[0086] S5-3. Solve the bipartite graph according to the Hungarian algorithm. The edge solved at this time is the match between the driver and the passenger. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
[0087] Furthermore, step S6 is specifically as follows:
[0088] S6-1. Find a set of priority matching passengers, and give priority to matching passengers with a high historical waiting time. Specifically, sort the passengers in descending order according to c(pi), select the passengers in the first 1 / 8 quartile to enter the priority matching passenger set Sp. If the passenger has been matched in the first stage, remove it from the set;
[0089] S6-2. Construct the edge that preferentially matches the neighbors of passengers. According to the set Sp, find the short neighbor set Neighbourp(pi) for each passenger. The neighbors are defined as those within the range of 3 kilometers of longitude and latitude locp and locd. Here, we need to ensure that these passengers can get a taxi. The distance is set to 3 kilometers. For the neighbor set Neighbourp(pi) of passenger pi, if dj has been matched in the first stage, remove dj from Neighbourp(pi), and construct an edge from pi to dj for graph G. The edge value is f(pi,dj), where dj∈Neighbourp(pi);
[0090] S6-3. Solve the bipartite graph according to the Hungarian algorithm. The edge solved at this time is the match between the passenger and the driver. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
[0091] Further, step S7 is specifically as follows:
[0092] S7-1. Find the driver set USd that has not been matched yet;
[0093] S7-2. Construct the edge of the driver's short neighbors. According to the set USd, find the short neighbor set MiddleNeighbourd(dj) for each driver. The neighbor is defined as the latitude and longitude locp and locd within a range of 2 kilometers. For the neighbor set MiddleNeighbourd(dj) of driver dj, if pi has been matched in the previous stage, remove pi from MiddleNeighbourd(dj) and construct an edge from pi to dj for graph G. The edge value is f(pi,dj), where pi∈MiddleNeighbourd(dj);
[0094] S7-3. Solve the bipartite graph according to the Hungarian algorithm. The edge solved at this time is the match between the driver and the passenger. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
[0095] Further, step S8 is specifically as follows:
[0096] S8-1, find the driver set VSd that has not been matched yet;
[0097] S8-2. Construct the edge of the driver's long neighbors. According to the set VSd, find the short neighbor set Neighbourd(dj) for each driver. The neighbors are defined as those within the range of 5 kilometers of the longitude and latitude locp and locd. For the neighbor set Neighbourd(dj) of driver dj, if pi has been matched in the previous stage, remove pi from Neighbourd(dj) and construct an edge from pi to dj for graph G. The edge value is f(pi,dj), where pi∈Neighbourd(dj);
[0098] S8-3. Solve the bipartite graph according to the Hungarian algorithm. The edge solved at this time is the match between the driver and the passenger; remove other unmatched edges in the graph G. At this time, the edges in the graph G are the result set R, and only the matching set solved in the fourth stage is recorded as Sl.
[0099] Although 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 appended claims and their equivalents.
Claims
1. A driver-passenger matching satisfaction method based on a multi-stage bipartite graph, characterized by: The specific steps include: S1. Optimization problem to establish a passenger satisfaction model; S2, optimization problem to establish a satisfaction model for drivers; S3. Optimization problem sets the optimization objective function; S4. Initialize the bipartite graph G: Initially, there are n+m vertices, pi, dj, 1≤i≤n, 1≤j≤m. Initially, there are no edge connections between vertices. S5, in the first stage of the bipartite graph, drivers with high historical driving time and high pick-up time are prioritized for matching; S6, the second stage of the bipartite graph, prioritizes matching passengers with historically high waiting times; S7, the third stage of the bipartite graph, the mid-range driver and passenger satisfaction matching; S8, the fourth stage of the bipartite graph, remote drivers and passengers’ satisfaction matching; S9. Improve passenger and driver satisfaction for long-distance orders: For the remote distance set Sl, check for matching orders with a pickup distance exceeding γ, where γ = 3.5 kilometers. Adjust the passenger price and driver commission rate for these orders to improve satisfaction and reduce cancellation rates. Step S1 is specifically as follows: S1-1. Let n passengers be assigned values from p1 to pn, and their geographical locations and longitudes be denoted as locp1 to locpn; let m drivers be assigned values from d1 to dm, and their geographical locations and longitudes be denoted as locd1 to locdm; f(pi,dj) is the estimated pick-up time for passenger pi and driver dj, in minutes, where 1≤i≤n, 1≤j≤m; hp(pi) is the satisfaction of the passenger pi who took the taxi, 1≤i≤n; S1-2. Calculate hp(pi) according to the formula: hp(pi)=(1-c(pi)) / (1+c(pi))×0.5+MAX((α-tp(pi)) / (α+tp(pi)),0)×0.5; where, tp(pi) is the time it takes to pick up the passenger after the driver is finally assigned, in minutes, 1≤i≤n; c(pi) is the percentage of long-duration taxi calls by the passenger in history, c(pi) = number of calls lasting more than 5 minutes by the passenger in the past 180 days / total number of taxi calls by the passenger in the past 180 days, 1≤i≤n; α = e × today's average pickup time. If no vehicle is assigned, tp(pi) = α, where e is a mathematical constant. Step S2 is specifically as follows: S2-1. Define hd(dj) as the satisfaction of driver dj in this taxi trip, 1≤j≤m; S2-2. Calculate hd(dj) according to the formula: hd(dj)=(1-g(dj)) / (1+g(dj))×0.5+MAX((β-td(dj)) / (α+td(dj)),0)×0.5; among them, g(dj) = (today's idle time + today's pick-up time) / number of orders received; td(dj) is the pick-up time of the driver after the final driver assignment, in minutes, 1≤j≤m; β = e × (today's average pick-up time + today's average idle time). If no passenger is assigned, td(dj) = β, where e is a mathematical constant. Step S3 is as follows: The optimization function L = ∑hp(pi) + ∑hd(dj), 1≤i≤n, 1≤j≤m. The problem is transformed into finding the matching set R and obtaining the maximum value of the function L; where (pi, dj) represents a matching pair in the R set, indicating the matching of orders pi and dj.
2. The driver-passenger matching satisfaction method based on a multi-stage bipartite graph according to claim 1, characterized in that: Step S5 is specifically as follows: S5-1. Find a priority matching driver set, prioritizing drivers with high idle rates and high pick-up times. Specifically, sort the drivers in descending order of g(dj) and select the top 1 / 8 quantile of drivers to enter the priority matching driver set Sd. S5-2. Construct edges that prioritize matching drivers' short neighbors. Based on the set Sd, find the short neighbor set ShortNeighbourd(dj) for each driver. Neighbors are defined as those within a 1-kilometer range of latitude and longitude locp and locd. For driver dj's short neighbor set ShortNeighbourd(dj), construct an edge from pi to dj on the graph G, with the edge value f(pi,dj), where pi∈ShortNeighbourd(dj). S5-3. Solve the bipartite graph using the Hungarian algorithm. The edge solved at this time is the match between the driver and the passenger. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
3. The driver-passenger matching satisfaction method based on a multi-stage bipartite graph according to claim 1, characterized in that: Step S6 is specifically as follows: S6-1. Find a set of priority matching passengers, prioritizing passengers with historically high waiting times. Specifically, sort the passengers in descending order by c(pi), select the passengers in the top 1 / 8 quartile and add them to the priority matching passenger set Sp. If a passenger has already been matched in the first stage, remove it from the set. S6-2. Construct edges that prioritize matching passengers' neighbors. Based on the set Sp, find the short neighbor set Neighbourp(pi) for each passenger. Neighbors are defined as those within a 3-kilometer range of longitude and latitude locp and locd. Here, to ensure that these passengers can get a taxi, the distance is set to 3 kilometers. For passenger pi's neighbor set Neighbourp(pi), if dj has been matched in the first stage, remove dj from Neighbourp(pi). Construct an edge from pi to dj on the graph G, with the edge value f(pi,dj), where dj∈Neighbourp(pi). S6-3. Solve the bipartite graph using the Hungarian algorithm. The edge solved at this time is the match between the passenger and the driver. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
4. The driver-passenger matching satisfaction method based on a multi-stage bipartite graph according to claim 1, characterized in that: Step S7 is specifically as follows: S7-1. Find the unmatched driver set USd; S7-2. Construct the short neighbor edges of the drivers. Based on the set USd, find the short neighbor set MiddleNeighbourd(dj) for each driver. Neighbors are defined as those within a 2-kilometer range of the longitude and latitude locp and locd. For the neighbor set MiddleNeighbourd(dj) of driver dj, if pi has been matched in the previous stage, remove pi from MiddleNeighbourd(dj) and construct an edge from pi to dj on the graph G. The edge value is f(pi,dj), where pi∈MiddleNeighbourd(dj). S7-3. Solve the bipartite graph using the Hungarian algorithm. The edge solved at this time is the match between the driver and the passenger. At the same time, for the subsequent algorithm, remove other unmatched edges in graph G.
5. The driver-passenger matching satisfaction method based on a multi-stage bipartite graph according to claim 1, characterized in that: Step S8 is specifically as follows: S8-1. Find the driver set VSd that has not yet been matched; S8-2. Construct the long neighbor edges of the drivers. Based on the set VSd, find the short neighbor set Neighbourd(dj) for each driver. Neighbors are defined as those within a 5-kilometer range of the longitude and latitude locp and locd. For driver dj's neighbor set Neighbourd(dj), if pi has been matched in the previous stage, remove pi from Neighbourd(dj). Construct an edge from pi to dj on the graph G with the edge value f(pi,dj), where pi∈Neighbourd(dj). S8-3. Solve the bipartite graph using the Hungarian algorithm. The edges solved at this time are the matches between the driver and the passenger. Remove other unmatched edges in graph G. At this time, the edges in graph G are the result set R. Only the matching set solved in the fourth stage is recorded as Sl.
Citation Information
Patent Citations
Optimal multi-vehicle scheduling method based on active distribution type taxi service system
CN106448138A
Online car-hailing intelligent order distribution method based on multi-objective optimization
CN109784523A
