A dynamic large-scale carpooling route planning method based on robust path
Through the dynamic large-scale joint route planning method based on robust paths, the problems of low success rate of joint match and insufficient real-time algorithm in the existing technology are solved, and the success rate of joint match and the efficiency of road network service is achieved. Effect.
Patent Information
- Application Number
- CN202111073090.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-09-14
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2041-09-14
AI Technical Summary
The existing technology cannot guarantee a high success rate of matching and matching, and the travel time window between drivers and passengers is not fully analyzed, and the algorithm is insufficient in large-scale matching scenarios, resulting in low dynamic matching and matching efficiency.
Using a dynamic large-scale route planning method based on robust paths, by establishing a data acquisition module and a driver-passenger dynamic match and path optimization model, filtering potential matching passengers, decomposing the optimal matching problem, and calculating the cost of travel inconvenient travel , choose a robust path to increase the chances of drivers picking up passengers, improve matching rates and algorithm real-time performance.
It improves the success rate of dynamic multiplication matching and the real-time algorithm, increases the opportunity for drivers to pick up and drop off passengers, and improves the efficiency of road network service and order matching rate.
Smart Images

Figure CN113962434B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of intelligent traffic management and intelligent control, and specifically relates to a dynamic large-scale carpooling route planning method based on robust paths, which is used to improve the efficiency of dynamic carpooling services in a road network. Background Art
[0002] Ridepooling is a mode of transportation in which travelers with similar travel routes share a vehicle. In most ridepooling problems, drivers and passengers are subject to travel time window constraints. A driver can visit different locations to pick up and drop off multiple passengers while meeting the travel time window constraints. The flexibility of the departure and arrival times of drivers and passengers represents the degree to which drivers and passengers are willing to take detours and is an important factor affecting the matching rate. Therefore, the dynamic large-scale ridepooling route planning method based on robust paths can study the uncertainty of driver travel time, increase the opportunity for drivers to pick up and drop off more passengers, and improve the successful matching rate of orders within the road network.
[0003] Through searching, it is found that Chinese patent document CN111932040A discloses a method and system for planning a carpooling route. The starting point and end point sent by the passenger terminal are connected to the nearby streets, and the time required for the passenger to reach the boarding station and the time for the driver to arrive at the boarding station are calculated. After comparison, the default order of passengers boarding the vehicle is obtained. The default route is pushed to the mobile terminal of the target vehicle, which is the optimized carpooling route. Chinese patent document CN111667086A discloses a method and system for optimizing vehicle carpooling routes. By establishing a carpooling matching and path optimization model with the highest driver income, the shortest vehicle driving distance, the least vehicle idle time and the highest passenger satisfaction, based on the real-time data of vehicles and passengers temporarily stored in the server module, the improved genetic algorithm and simulated annealing algorithm are used to solve the multi-vehicle carpooling matching and path optimization model, and finally the carpooling matching of drivers and passengers and the planning of routes are obtained. Chinese patent document CN112418973A discloses a method and system for processing carpooling order data on a big data online car-hailing platform, which determines whether the first passenger and the second passenger are on the same route through the first vector, the second vector and the third vector. Among them, the starting point information of the first carpooling order points to the end point information as the first vector, the starting point information of the second carpooling order points to the end point information as the second vector, and the starting point information of the first carpooling order points to the starting point information of the second carpooling order as the third vector. When the cosine value of the first vector and the second vector is greater than or equal to 0, and the cosine value of the first vector and the third vector is greater than or equal to 0, the estimated travel time from the starting point to the end point of the first carpooling order and the estimated travel time to complete the entire journey of the first carpooling order and the second carpooling order are calculated. Determine whether the difference between the two travel times is less than the threshold value. If so, the passengers of the first carpooling order and the second carpooling order are on the same route, and the carpooling is successful.
[0004] However, the above technical solution has the following shortcomings:
[0005] First, it cannot guarantee a high carpooling matching success rate. The carpooling matching success rate represents the efficiency of the road network service. The above technical solution can achieve the matching of drivers and passengers and route planning, but it cannot guarantee a high matching rate.
[0006] Second, the travel time windows of drivers and passengers are not fully analyzed. Most of the above technical solutions use the travel time window as a constraint to screen the matching data, but do not consider the impact of the uncertainty of travel time on the driver's idle rate, road network service rate, etc.
[0007] Third, in large-scale carpooling scenarios, the real-time performance of the algorithm cannot be guaranteed. Since new order information will enter the system and wait for allocation every time the system is updated, how to reduce the search time and achieve efficient dynamic carpooling matching is the research focus of dynamic carpooling.
[0008] In summary, how to improve the matching success rate of dynamic carpooling and the real-time performance of the algorithm is a technical problem that technical personnel in this field urgently need to solve. Summary of the invention
[0009] In view of the shortcomings of the prior art, the present invention proposes a dynamic large-scale carpooling route planning method based on robust paths to improve the dynamic carpooling matching success rate and the real-time performance of the algorithm in the road network. The present invention specifically adopts the following technical solutions:
[0010] A dynamic large-scale carpooling route planning method based on robust path, the method comprising the following steps:
[0011] Step 1: Create a data acquisition module
[0012] The data collection module is used to collect the driver and passenger data currently input into the system, and record the latitude and longitude coordinates of the starting and ending points of all data, the request time, the expected departure time, the expected arrival time, and the tolerance time value attributes of departure and arrival;
[0013] Step 2: Build a driver-passenger dynamic ride-sharing matching and route optimization model
[0014] Step 2-1: Establish the constraints of the dynamic carpooling matching and route optimization model. The specific constraints are as follows:
[0015]
[0016]
[0017]
[0018]
[0019]
[0020]
[0021]
[0022]
[0023]
[0024]
[0025] Among them, a dr ,y r , and is a decision variable. If driver d serves passenger r, a dr is 1; if passenger r is not served by any driver, y r is 1; if driver d picks up a passenger at road section (i, j), is 1; if driver d travels through path (i,j), is 1;
[0026] A represents the set of road segments in the road network, D represents the set of drivers, R represents the set of passengers, t ij represents the travel time from node i to node j, o d represents the starting point of driver d, d d represents the destination of driver d, o r represents the starting point of passenger r, d r represents the destination of passenger r, t d represents the maximum travel time that driver d is willing to spend, q d represents the maximum number of people that driver d’s vehicle can carry, represents the earliest time that driver d can depart, t od represents the latest time that driver d can depart, represents the earliest time that driver d can reach the destination, represents the latest time that driver d can arrive at the destination, represents the earliest time that passenger r can depart, represents the latest time that passenger r can depart, represents the earliest time that passenger r can arrive at the destination, represents the latest time that passenger r can arrive at the destination, represents the actual boarding time of passenger r, represents the estimated arrival time of passenger r;
[0027] Step 2-2: Establish a dynamic carpooling matching and route optimization model
[0028]
[0029] in, represents the information release time of passenger r, p r represents the penalty value for passenger r not being served by any driver;
[0030] Our step 3: optimal driver-passenger matching and shortest path planning
[0031] Step 3-1: Based on the real-time data of drivers and passengers in the road network recorded by the data collection module in step 1, all passengers that can potentially match each driver d are screened out through constraint conditions;
[0032] Step 3-2: Decompose the optimal matching problem of all drivers in the road network into independent sub-problems for optimal matching. Sort the passengers that each driver d can potentially match. Each sort is a pick-up and drop-off matching plan. According to the dynamic carpooling matching and path planning model, traverse all the arrangement plans of each driver picking up and dropping off passengers, calculate the travel inconvenience cost of all plans, and obtain the optimal driver-passenger matching and path planning after comparison. The planned path is the shortest path and is stored;
[0033] Step 4: Create a set of drivers’ uncertain travel times
[0034] Step 4-1: Define the uncertainty level Γ, which represents the deviation level of the travel time of the remaining paths relative to the shortest path. Γ is expressed as follows:
[0035]
[0036]
[0037] Among them, A d is a set of robust paths subject to travel time uncertainty, A d ∈A; represents the deviation rate of the robust path m relative to the shortest path, M ij Represents the set of all path number indexes from node i to node j that satisfy the time window constraint;
[0038] Step 4-2: Define the set of drivers’ uncertain travel times
[0039]
[0040] in, is the nominal value of the travel time, which indicates the travel time of driver d from node i to node j by taking the shortest route. is the latest time that driver d can accept to arrive at the destination, represents the estimated arrival time of driver d, z is the deviation rate vector of all robust paths between two nodes;
[0041] Step 5: Robust path selection during dynamic carpooling
[0042] Step 5-1: Calculate the range of driver travel time
[0043] Based on the shortest path for driver d obtained in step 3, calculate the estimated arrival time of driver d at destination d by taking the shortest path. d time, calculate the latest time that driver d is willing to arrive at the destination The time it takes to reach the destination by taking the shortest route with driver D The difference between
[0044] Step 5-2: Calculate the travel time fluctuation value and fluctuation rate of each robust path
[0045] The calculation method of the travel time fluctuation value of each robust path is: the travel time of the robust path minus the shortest path travel time;
[0046] The calculation method of the travel time fluctuation rate of each robust path is: the ratio of the fluctuation value of the robust path to the maximum fluctuation value of the current driver d’s travel time;
[0047] Step 5-3: Set the uncertainty level Γ and select the optimal robust path
[0048] The method of dynamic fluctuation travel time is adopted to make the fluctuation value of the robust path selected between nodes less than or equal to the maximum fluctuation value of the current driver d's travel time. By setting the uncertainty level Γ, the traversal driver d travels from the current node to the end point d. d All robust paths of are obtained to obtain a set of paths whose cumulative volatility of travel time is close to or equal to the uncertainty level Γ. This set of robust paths is the optimized robust route.
[0049] Step 6: After the driver and the passenger complete their trip, all the routes they have taken are recorded respectively.
[0050] The present invention has the following beneficial effects:
[0051] The present invention establishes a driver-passenger dynamic carpooling matching and path optimization model. When solving the problem, the optimal matching problem of all drivers in the road network is decomposed into independent sub-problems for optimal matching, thereby improving the real-time performance of the algorithm. According to the optimization model, all arrangement schemes of the driver picking up and dropping off passengers are traversed to obtain the driver-passenger optimal matching and shortest path planning. Based on the shortest path for the driver to pick up and drop off passengers, the decision maker can calculate the fluctuation range of the driver's travel time according to the driver's travel time window. By setting the uncertainty level Γ, a robust path is selected by dynamically fluctuating the driver's travel time, which increases the driver's chance of picking up and dropping off more passengers, thereby improving the order matching rate and increasing the road network service level. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] Figure 1 The present invention is a flow chart of the method.
[0053] Figure 2 This is a flow chart of the driver-passenger optimal matching and shortest path planning method of the present invention.
[0054] Figure 3 Flowchart of the robust path planning method for driver-passenger. DETAILED DESCRIPTION
[0055] The present invention discloses a method for dynamic large-scale carpooling route planning based on a robust path, and the specific steps are as follows:
[0056] Step 1: Establish a data collection module. The data collection module is responsible for collecting the driver and passenger data currently input into the system, and recording all data attributes such as the latitude and longitude coordinates of the starting and ending points, the request time, the expected departure time, the expected arrival time, and the tolerance time value for departure and arrival.
[0057] Step 2: Establish a driver-passenger dynamic carpooling matching and route optimization model.
[0058] Step 2-1: Establish constraints for the dynamic carpooling matching and route optimization model. Through the constraints, we can filter out orders that meet the driver's pick-up time window and other conditions, that is, the set of passengers that each driver can potentially match, thereby reducing the scale of the road network and improving the real-time performance of the dynamic carpooling algorithm. The specific constraints are as follows:
[0059]
[0060]
[0061]
[0062]
[0063]
[0064]
[0065]
[0066]
[0067]
[0068]
[0069] Among them, (1) ensure that a passenger can only be served by one driver; (2) ensure that the total number of passengers in a vehicle cannot exceed the maximum capacity of the vehicle; (3) and (4) ensure that the driver departs and arrives within the travel time window; (5)-(8) ensure that if a passenger is served, he must depart and arrive within his travel time window; (9) ensure that the driver's last arrival location is the driver's destination; (10) ensure that the driver's total travel time cannot be greater than the maximum travel time the driver is willing to spend.
[0070] a dr ,y r , and is a decision variable. If driver d serves passenger r, a dr is 1; if passenger r is not served by any driver, y r is 1; if driver d picks up a passenger at road section (i, j), is 1; if driver d travels through path (i,j), is 1.
[0071] A represents the set of road segments in the road network, D represents the set of drivers, and R represents the set of passengers. ij represents the travel time from node i to node j, o d represents the starting point of driver d, d d represents the destination of driver d, o r represents the starting point of passenger r, d r represents the destination of passenger r, t d represents the maximum travel time that driver d is willing to spend, q d represents the maximum number of people that driver d’s vehicle can carry, represents the earliest time that driver d can depart, t od represents the latest time that driver d can depart, represents the earliest time that driver d can reach the destination, represents the latest time that driver d can arrive at the destination, represents the earliest time that passenger r can depart, represents the latest time that passenger r can depart, represents the earliest time that passenger r can arrive at the destination, represents the latest time that passenger r can arrive at the destination, represents the actual boarding time of passenger r, represents the estimated arrival time of passenger r,
[0072] Step 2-2: Establish a dynamic carpooling matching and route optimization model, with the optimization goals of minimizing driver travel time, minimizing passenger waiting time for boarding, minimizing passenger delay time at the destination, and minimizing the penalty for unmatched passengers.
[0073]
[0074] in, represents the information release time of passenger r, p r represents the penalty value for passenger r not being served by any driver.
[0075] Step 3: Achieve optimal driver-passenger matching and shortest path planning.
[0076] Step 3-1: Based on the real-time data of drivers and passengers in the road network recorded by the data collection module in step 1, all passengers that can potentially match each driver d are screened out through constraint conditions.
[0077] Step 3-2: Decompose the optimal matching problem of all drivers in the road network into independent sub-problems for optimal matching, and sort the passengers that each driver d can potentially match. Each sort is a pick-up and drop-off matching plan. According to the dynamic carpooling matching and path planning model, traverse all the arrangement plans for each driver to pick up and drop off passengers, calculate the travel inconvenience cost of all plans, and obtain the optimal driver-passenger matching and path planning after comparison. The planned path is the shortest path and is stored.
[0078] Step 4: Establish a set of drivers’ uncertain travel times.
[0079] Step 4-1: Define the uncertainty level Γ, which represents the deviation level of the travel time of the remaining paths relative to the shortest path.
[0080] Among them, the remaining paths that have deviations from the shortest path travel time are called robust paths. Γ is expressed as follows:
[0081]
[0082]
[0083] Among them, A d is a set of robust paths subject to travel time uncertainty, A d ∈A. M represents the deviation rate of the robust path m relative to the shortest path. ij Represents the set of indexes of all paths from node i to node j that satisfy the time window constraint. The path numbers are sorted in ascending order of the offset rate. It means that the decision maker makes decisions to a conservative degree that does not choose the shortest path solution.
[0084] Step 4-2: Define the set of drivers’ uncertain travel times.
[0085]
[0086] in, is the nominal value of the travel time, which indicates the travel time of driver d from node i to node j by taking the shortest route. is the latest time that driver d can accept to arrive at the destination, represents the estimated arrival time of driver d, and z is the deviation rate vector of all robust paths between two nodes.
[0087] Step 5: Selection of robust paths in the dynamic carpooling process. Based on the shortest path of driver d obtained in step 3, the method of dynamically fluctuating travel time is used to select a robust path with a longer travel time relative to the shortest path, so as to moderately increase the driver's travel time and increase the opportunity to pick up and drop off passengers.
[0088] Step 5-1: Calculate the range of fluctuation of driver travel time. Based on the shortest path obtained in step 3, driver d travels from the current node to the end point d. d The shortest set of sections passed, calculate the expected arrival time of driver d by taking the shortest route d Calculate the latest time that driver d is willing to arrive at the destination The time it takes to reach the destination by taking the shortest route with driver D The difference between the current driver d's travel time and the maximum value of the fluctuation is As the fluctuation range of the current driver d’s travel time.
[0089] Step 5-2: Calculate the travel time fluctuation value and fluctuation rate of each robust path. The calculation method of the travel time fluctuation value of each robust path is: the travel time of the robust path minus the travel time of the shortest path. The calculation method of the travel time fluctuation rate of each robust path is: the ratio of the fluctuation value of the robust path to the maximum fluctuation value of the current driver d's travel time.
[0090] Step 5-3: Set the uncertainty level Γ and select the optimal robust path. When the driver d's travel time fluctuation range is satisfied, the decision maker can choose a robust path. The dynamic fluctuation travel time method is used to make the fluctuation value of the robust path selected between nodes less than or equal to the maximum fluctuation value of the current driver d's travel time, that is, the cumulative fluctuation rate of the robust path selected between nodes is less than or equal to 1. By setting the uncertainty level Γ, the traversal driver d travels from the current node to the end point d. d All robust paths of are obtained to obtain a set of paths whose cumulative volatility of travel time is close to or equal to the uncertainty level Γ. This set of robust paths is the optimized robust route.
[0091] Step 6: After the driver and the passenger complete their trip, all the routes they have taken are recorded respectively.
Claims
1. A dynamic large-scale carpooling route planning method based on robust path, characterized in that: The method comprises the following steps: Step 1: Create a data acquisition module The data collection module is used to collect the driver and passenger data currently input into the system, and record the latitude and longitude coordinates of the starting and ending points of all data, the request time, the expected departure time, the expected arrival time, and the tolerance time value attributes of departure and arrival; Step 2: Build a driver-passenger dynamic ride-sharing matching and route optimization model Step 2-1: Establish the constraints of the dynamic carpooling matching and route optimization model. The specific constraints are as follows: Among them, a dr ,y r , and is a decision variable. If driver d serves passenger r, a dr is 1; if passenger r is not served by any driver, y r is 1; if driver d picks up a passenger at road section (i, j), is 1; if driver d travels through path (i,j), is 1; A represents the set of road segments in the road network, D represents the set of drivers, R represents the set of passengers, t ij represents the travel time from node i to node j, o d represents the starting point of driver d, d d represents the destination of driver d, o r represents the starting point of passenger r, d r represents the destination of passenger r, t d represents the maximum travel time that driver d is willing to spend, q d represents the maximum number of people that driver d’s vehicle can carry, represents the earliest time that driver d can depart, t od represents the latest time that driver d can depart, represents the earliest time that driver d can reach the destination, represents the latest time that driver d can arrive at the destination, represents the earliest time that passenger r can depart, represents the latest time that passenger r can depart, represents the earliest time that passenger r can arrive at the destination, represents the latest time that passenger r can arrive at the destination, represents the actual boarding time of passenger r, represents the estimated arrival time of passenger r; Step 2-2: Establish a dynamic carpooling matching and route optimization model in, represents the information release time of passenger r, p r Indicates the penalty value for passenger r not being served by any driver; Step 3: Optimal driver-passenger matching and shortest path planning Step 3-1: Based on the real-time data of drivers and passengers in the road network recorded by the data collection module in step 1, all passengers that can potentially match each driver d are screened out through constraint conditions; Step 3-2: Decompose the optimal matching problem of all drivers in the road network into independent sub-problems for optimal matching. Sort the passengers that each driver d can potentially match. Each sort is a pick-up and drop-off matching plan. According to the dynamic carpooling matching and path planning model, traverse all the arrangement plans of each driver picking up and dropping off passengers, calculate the travel inconvenience cost of all plans, and obtain the optimal driver-passenger matching and path planning after comparison. The planned path is the shortest path and is stored; Step 4: Create a set of drivers’ uncertain travel times Step 4-1: Define the uncertainty level Γ, which represents the deviation level of the travel time of the remaining paths relative to the shortest path. Γ is expressed as follows: Among them, A d is a set of robust paths subject to travel time uncertainty, A d ∈A; represents the deviation rate of the robust path m relative to the shortest path, M ij Represents the set of all path number indexes from node i to node j that satisfy the time window constraint; Step 4-2: Define the set of drivers’ uncertain travel times in, is the nominal value of the travel time, which indicates the travel time of driver d from node i to node j by taking the shortest route. is the latest time that driver d can accept to arrive at the destination, represents the estimated arrival time of driver d, z is the deviation rate vector of all robust paths between two nodes; Step 5: Robust path selection during dynamic carpooling Step 5-1: Calculate the range of driver travel time Based on the shortest path for driver d obtained in step 3, calculate the estimated arrival time of driver d at destination d by taking the shortest path. d time, calculate the latest time that driver d is willing to arrive at the destination The time it takes to reach the destination by taking the shortest route with driver D The difference between Step 5-2: Calculate the travel time fluctuation value and fluctuation rate of each robust path The calculation method of the travel time fluctuation value of each robust path is: the travel time of the robust path minus the shortest path travel time; The calculation method of the travel time fluctuation rate of each robust path is: the ratio of the fluctuation value of the robust path to the maximum fluctuation value of the current driver d’s travel time; Step 5-3: Set the uncertainty level Γ and select the optimal robust path The method of dynamic fluctuation travel time is adopted to make the fluctuation value of the robust path selected between nodes less than or equal to the maximum fluctuation value of the current driver d's travel time. By setting the uncertainty level Γ, the traversal driver d travels from the current node to the end point d. d All robust paths of are obtained to obtain a set of paths whose cumulative volatility of travel time is close to or equal to the uncertainty level Γ. This set of robust paths is the optimized robust route. Step 6: After the driver and the passenger complete their trip, all the routes they have taken are recorded respectively.
