Dynamic multi-objective optimization method integrating bus timetable and vehicle scheduling
Through dynamic spatiotemporal network flow model and multi-objective optimization algorithm, and combining pre-order operation information to adjust the bus schedule and vehicle scheduling, the problems of insufficient information utilization and target conflict in the existing technology are solved, and the dynamic scheduling effect of the bus system is improved.
Patent Information
- Application Number
- CN202510531957.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-25
- Publication Date
- 2025-08-01
AI Technical Summary
The existing dynamic bus scheduling method fails to fully utilize the information from the previous operation period and fails to effectively handle the conflict between multiple optimization goals, resulting in a reduction in the overall effectiveness of the optimization results and it is difficult to obtain a global optimal solution.
Through dynamic spatiotemporal network flow model, the mapping relationship between operation status and operation plan is designed, combined with pre-order operation information, the quadratic polynomial function is used to calculate the probability of one-way connection arc selection, and the sampling and multi-objective optimization algorithm are used to generate a better timetable and vehicle scheduling integration solution.
It has achieved dynamic adjustment of bus schedules and vehicle scheduling according to changes in the operating environment, improved the quality of bus services and operational efficiency, and improved the reliability and operational level of the bus system.
Smart Images
Figure CN120409812A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of public transportation scheduling optimization, and particularly relates to a dynamic multi-objective optimization method integrating bus timetables and vehicle scheduling. Background Art
[0002] Buses are an important part of urban public transportation and guarantee for citizens' travel. Bus timetables and vehicle scheduling plans are important bases for bus operations and determine the service level of buses. However, during actual operations, unplanned interference events such as traffic congestion, bad weather, and vehicle failures often cause bus services to deviate from the established plan. This deviation between the plan and the execution seriously affects the service quality. If not adjusted in time, these deviations may gradually deteriorate, ultimately leading to the abnormal operation of bus lines, thereby reducing the reliability and operation efficiency of the public transportation system. Therefore, re-preparing the scheduling plan according to the actual situation for bus timetables and vehicle scheduling, also known as dynamic scheduling, is an important means to adapt to complex environments and improve the bus service level.
[0003] Most existing dynamic scheduling methods independently execute optimization strategies during each operation period. However, this method fails to fully utilize the information accumulated in the previous operation period, which could provide important guidance for the optimization search in the current period, thus preventing the algorithm from falling into local optimal solutions. At the same time, these studies usually use the weighted summation method to integrate multiple optimization objectives into a single objective function, without fully considering the inherent conflicts between the objectives, resulting in a reduction in the overall effectiveness of the optimization results and further increasing the difficulty of obtaining the global optimal solution. Therefore, against the background of the limitations of existing methods, this patent proposes a dynamic multi-objective optimization method integrating bus timetables and vehicle scheduling, which can combine the operation information of the previous operation period to generate a better integrated dynamic scheduling plan for timetables and vehicle scheduling, and at the same time consider three costs: the passenger travel cost, the vehicle operation cost, and the penalty cost, to improve the operation level and service quality of buses. Summary of the Invention
[0004] Aiming at the gap in the existing bus dynamic scheduling that does not consider the integrated scheduling of bus timetables and vehicle scheduling, the present invention proposes a dynamic multi-objective optimization method integrating bus timetables and vehicle scheduling. By modeling the bus scheduling problem through a dynamic spatio-temporal network flow model and designing the mapping relationship between the operation state and the operation plan, a dynamic multi-objective optimization method integrating bus timetables and vehicle scheduling that can simultaneously adjust the timetable and the vehicle scheduling plan according to changes in the operation environment is proposed. The specific steps are as follows: Step 1: Initialize the solution set. Given the number of vehicles and the operating time period, divide the operating time period of the bus into multiple time slots at 2-hour intervals. At the same time, generate a candidate one-way task every minute to form a dynamic spatio-temporal network for bus vehicle scheduling. Randomly form 100 initial solutions composed of one-way connection arcs based on the dynamic spatio-temporal network for bus vehicle scheduling to form the initial solution set; Step 2: Obtain the operating state. According to the operating data, obtain the operating state in the current time slot. The operating state includes the arrival rate of passengers at each stop of the line, the alighting rate of passengers, and the operating time of the candidate one-way trips during this time slot; where the arrival rate of passengers is the number of passengers arriving per minute, and the alighting rate of passengers is the ratio of the number of alighting passengers to the number of passengers in the vehicle when the bus arrives at the stop; Step 3: Calculate the selection probability of one-way connection arcs. The probability of whether each one-way connection arc is selected is defined as a quadratic polynomial function of the operating state. In order to make full use of the information in the previous time slot, the coefficients of the quadratic polynomial function are determined by the operating state at the previous moment and the current solution set using the least squares method. Combine the coefficients using the quadratic polynomial regression equation to calculate the selection probabilities of all one-way connection arcs according to the operating state at the current moment; Step 4: Sample to generate new solutions. Sample the candidate one-way connection arcs according to the selection probabilities of the one-way connection arcs to generate 10 new solutions, and randomly replace 10 solutions in the solution set; Step 5: Multi-objective optimization. Use the search operator to generate 30 new solutions, and then use the selection operator. Combine the non-dominated sorting method for multiple objectives to select 100 solutions from the current 100 solutions and 30 new solutions to form a new solution set; Step 6: Process different time slots in sequence until all time slots are processed. Description of the Drawings
[0005] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.
[0006] Figure 1 is a schematic flowchart of the method of an embodiment of the present invention; Figure 2 is a schematic diagram of the dynamic spatio-temporal network for bus vehicle scheduling of an embodiment of the present invention; Figure 3 is a schematic diagram of the solution of an embodiment of the present invention. Detailed Embodiment
[0007] To enable those skilled in the art to better understand the technical solutions in the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0008] As Figure 1 shown, the present invention provides a schematic flowchart of a method for an embodiment, including: Step 1: Initialize the solution set. Given the number of vehicles and the operation time period, divide the operation time period of the bus into multiple time periods at intervals of 2 hours. At the same time, generate a candidate one-way task every minute to form a dynamic spatio-temporal network for bus vehicle scheduling. Randomly form 100 initial solutions composed of one-way connection arcs according to the dynamic spatio-temporal network for bus vehicle scheduling to form an initial solution set; Step 2: Obtain the operation state. According to the operation data, obtain the operation state in the current time period. The operation state includes the arrival rate of passengers at each station on the line, the alighting rate of passengers, and the operation time of the candidate one-way trips within this time period; where the arrival rate of passengers is the number of passengers arriving per minute, and the alighting rate of passengers is the ratio of the number of alighting passengers to the number of passengers in the vehicle when the bus arrives at the station; Step 3: Calculate the selection probability of one-way connection arcs. The probability of each one-way connection arc being selected is defined as a quadratic polynomial function of the operation state. In order to make full use of the information in the previous time period, the coefficients of the quadratic polynomial function are determined by the operation state at the previous moment and the current solution set using the least squares method. Combine the coefficients using the quadratic polynomial regression equation and calculate the selection probabilities of all one-way connection arcs according to the operation state at the current moment; Step 4: Sample to generate a new solution. Sample the candidate one-way connection arcs according to the selection probabilities of the one-way connection arcs to generate 10 new solutions, and randomly replace 10 solutions in the solution set; specifically: Step 41: According to the selection probabilities of the one-way connection arcs, sample whether the selection probability of each one-way connection arc is selected to be added to the solution through the roulette method, and finally form a new solution; Step 42: Check whether there is a one-way trip in the new solution that serves as the starting point of multiple one-way connection arcs. If so, randomly retain one of the one-way connection arcs; Step 43: Check whether there is a one-way trip in the new solution that serves as the ending point of multiple one-way connection arcs. If so, randomly retain one of the one-way connection arcs; Step 44: Randomly select two vehicle trip chains and determine whether there is a candidate one-way connection arc between the starting one-way trip and the ending one-way trip of the two vehicle trip chains. If so, add it to the solution; Step 45: Repeat the above steps until 10 new solutions are generated; Step 5: Multi-objective optimization. Use the search operator to generate 30 new solutions, and then use the selection operator. Combine the non-dominated sorting method for multiple objectives to select 100 solutions from the current 100 solutions and the 30 new solutions to form a new solution set. Specifically: Step 51: Randomly select 30 current solutions to execute the search operator, where the search operator includes a random deletion operator, a vehicle deletion operator, and a random addition operator. The random deletion operator means randomly selecting a one-way trip, deleting the one-way connection arcs connecting the two one-way trips connected to it, and at the same time obtaining the two one-way trips connected to it and adding the one-way connection arc between the two connected one-way trips. The vehicle deletion operator means randomly selecting a train trip chain to delete. The random addition operator means randomly selecting a candidate one-way connection arc and adding it to the solution. Step 52: Execute the selection operator on the 30 new solutions formed by the search operator and the current 100 solutions. The selection operator includes first calculating the passenger travel cost, vehicle operation cost, and penalty cost corresponding to all solutions. Then, use the fast non-dominated sorting method to identify all non-dominated solutions from the solution set. According to the structural characteristics of the train trip chains in each non-dominated solution, perform diversity screening on them. Specifically, the operator will calculate the length difference between the shortest train trip chain and the longest train trip chain in each non-dominated solution, and select the solutions in the top 10% and bottom 10% with the largest difference to ensure the diversity of the train trip chain structure. For the solutions in the middle 80%, they will be selected to enter the solution set by random sampling until the solution set contains 100 solutions. Step 6: Process different time periods in sequence until all time periods are processed.
[0009] As Figure 2 shown, the present invention provides a schematic diagram of a dynamic spatio-temporal network for bus vehicle scheduling in an embodiment. The network schematic diagram includes three types of nodes, namely depot nodes, which represent the depots where buses are parked. Each bus must depart from the depot, complete a series of one-way trips, and then return to the depot. A one-way trip represents an operation task from the starting station to the terminal station. A one-way trip that has started execution means that the departure time of the one-way trip is before the start time of the current time period. A one-way trip that has not started execution means that the departure time of the one-way trip is after the start time of the current time period. If the arrival time of the previous one-way trip is earlier than the start time of the next one-way trip for any two one-way trips, there is a one-way connection arc between the two one-way trips, indicating that the vehicle can continue to execute the next one-way trip after completing the previous one-way trip. The one-way connection arc pointing to a one-way trip that has started execution is called a started connection arc, and the one-way connection arc pointing to a one-way trip that has not started execution is called an unstarted connection arc. In addition, there is an outbound arc between the depot and any one-way trip, indicating that the vehicle departs from the depot to perform the one-way trip task. There is a return arc between any one-way trip and the depot, indicating that the vehicle returns to the depot after completing the one-way trip.
[0010] It should be noted that the depot is known, while the one-way trips are not. Therefore, the one-way trips in the spatio-temporal network are generated according to the operation period. Specifically, within the operation period, a one-way trip task is generated every minute, and then one-way connection arcs are generated based on the relationship between the departure time and arrival time of one-way trips.
[0011] As Figure 3 shown, the present invention provides a schematic diagram of the solution of an embodiment. Specifically, the solution consists of a set of selection variables of one-way connection arcs, and each variable x ij = 1 indicates that the connection arc between one-way trip i and one-way trip j is selected, otherwise x ij = 0. When converting the solution into a vehicle scheduling plan and a timetable, first, a set of train number chains need to be obtained from the solution. A train number chain is composed of a set of sequentially connected one-way trips. When obtaining the train number chain, first select a one-way trip i , and for any one-way trip i not equal to j , if x ij = 1 and x ji = 0, then take the one-way trip i as the starting one-way trip of the train number chain, and then find the one-way trip connected to it as the subsequent one-way trip. Repeat the one-way trip selection operation until there is no subsequent connected one-way trip. At this time, connect the depot to the starting one-way trip i , and then connect the last one-way trip to the depot to form a train number chain. Repeat the process of obtaining the train number chain from the solution until all one-way trips connected by one-way connection arcs are included in the train number chains. At this time, the set of all train number chains constitutes a vehicle scheduling plan, and the departure times of all one-way trips in the train number chains together constitute the departure timetable.
[0012] Although the present invention has been described in detail by referring to the accompanying drawings and in combination with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, those of ordinary skill in the art can make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions should all be within the scope covered by the present invention. Or any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the protection scope of the claims.
Claims
1. A dynamic multi-objective optimization method integrating bus timetables and vehicle scheduling, characterized in that It can solve the bus departure schedule and vehicle dispatching plan simultaneously according to historical operation information. The steps are as follows: Step 1: Initialize the solution set. Given the number of vehicles and the operation time period, the operation time period of the bus is divided into multiple time slots at 2-hour intervals. At the same time, a candidate one-way task is generated every minute to form a dynamic space-time network for bus vehicle dispatching. 100 initial solutions composed of one-way connection arcs are randomly formed according to the dynamic space-time network for bus vehicle dispatching to form an initial solution set; Step 2: Obtain the operation state. According to the operation data, obtain the operation state in the current time slot. The operation state includes the arrival rate of passengers at each stop of the line, the alighting rate of passengers, and the operation time of candidate one-way trips in this time slot; Among them, the arrival rate of passengers is the number of passengers arriving per minute, and the alighting rate of passengers is the ratio of the number of alighting passengers to the number of passengers in the vehicle when the bus arrives at the stop; Step 3: Calculate the selection probability of one-way connection arcs. The probability of each one-way connection arc being selected is defined as a quadratic polynomial function of the operation state. In order to make full use of the information in the previous time slot, the coefficients of the quadratic polynomial function are determined by the least squares method using the operation state and solution set at the previous moment. Combine the coefficients using the quadratic polynomial regression equation to calculate the selection probability of all one-way connection arcs according to the operation state at the current moment; Step 4: Sample to generate new solutions. Sample the candidate one-way connection arcs according to the selection probability of one-way connection arcs to generate 10 new solutions, and randomly replace 10 solutions in the solution set; Step 5: Multi-objective optimization. Use the search operator to generate 30 new solutions, and then use the selection operator. Combine the non-dominated sorting method for multiple objectives to select 100 solutions from the current 100 solutions and 30 new solutions to form a new solution set; Step 6: Process different time slots in turn until all time slots are processed.
2. The method according to claim 1, wherein In the sampling to generate new solutions, sample the candidate one-way trips according to the selection probability of one-way connection arcs to generate 10 new solutions, and randomly replace 10 solutions in the solution set, which is characterized by the following steps: Step 41: According to the selection probability of one-way connection arcs, sample whether the selection probability of each one-way connection arc is selected and added to the solution through the roulette wheel method, and finally form a new solution; Step 42: Check whether there is a one-way trip in the new solution that serves as the starting point of multiple one-way connection arcs. If so, randomly retain one of the one-way connection arcs; Step 43: Check whether there is a one-way trip in the new solution that serves as the end point of multiple one-way connection arcs. If so, randomly retain one of the one-way connection arcs; Step 44: Randomly select two train chains, and judge whether there is a candidate one-way connection arc between the starting one-way trip and the ending one-way trip of the two train chains. If so, add it to the solution; Step 45: Repeat the above steps until 10 new solutions are generated.
3. The method according to claim 1, wherein for the multi-objective optimization, 30 new solutions are generated by using a search operator, and then a selection operator is used to select 100 solutions from the current 100 solutions and the 30 new solutions to form a new solution set in combination with the non-dominated sorting method for multiple objectives, characterized in that It includes the following steps: Step 51: Randomly select 30 current solutions to execute the search operator, where the search operator includes a random deletion operator, a vehicle deletion operator, and a random addition operator; Among them, the random deletion operator means randomly selecting a one-way trip, deleting the two one-way connection arcs connecting it, obtaining the two one-way trips connected to it at the same time, and adding the one-way connection arc between the two connected one-way trips; the vehicle deletion operator means randomly selecting a train trip chain to delete; the random addition operator means randomly selecting a candidate one-way connection arc and adding it to the solution. Step 52: Execute the selection operator on the 30 new solutions formed by the search operator and the current 100 solutions; among them, the selection operator includes calculating the passenger travel cost, vehicle operation cost, and penalty cost corresponding to all solutions first; then, using the fast non-dominated sorting method to identify all non-dominated solutions from the solution set; according to the structural characteristics of the train trip chains in each non-dominated solution, perform diversity screening on them; specifically, the operator will calculate the length difference between the shortest train trip chain and the longest train trip chain in each non-dominated solution, and select the solutions in the top 10% and bottom 10% with the largest difference to ensure the diversity of the train trip chain structure; for the solutions in the middle 80%, they are selected to enter the solution set through random sampling until the solution set contains 100 solutions.