Metro crew scheduling management system and method based on MTSP
By transforming the subway flight scheduling problem into multi-travel merchant problems, using a hybrid method of simulated annealing and genetic algorithms, optimizing the flight attendant segment connection time and the workload of drivers and crew members, generating the optimal flight scheduling plan, solving the existing technology The problems of low efficiency in preparing flight attendants' scheduling schedules and unbalanced tasks have improved operational efficiency and driver and crew efficiency.
Patent Information
- Application Number
- CN202411501283.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2024-11-26
AI Technical Summary
It is difficult for the existing technology to quickly prepare efficient flight attendant scheduling plans, resulting in unbalanced task volume and low efficiency of drivers and crew members, which cannot meet the needs of urban rail transit operation.
The subway flight attendant scheduling problem is transformed into a multi-travel merchant problem (MTSP). Through a hybrid method combining simulated annealing algorithm and genetic algorithm, the flight attendant segment connection time and the workload balance of drivers and crew members are optimized to generate the optimal flight attendant scheduling plan.
The number of flight attendant tasks and the length of flight attendant segments have been significantly optimized, the labor costs of enterprises have been reduced, the duty-call efficiency of drivers and crew members has been improved, and the efficiency and quality problems of manual crew scheduling plans have been solved.
Smart Images

Figure CN119026876A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of digital management, and in particular to a subway crew scheduling management system and method based on MTSP. Background Art
[0002] The subway crew plan is the core document guiding the daily work of drivers and crew members, and can be divided into crew scheduling plans and crew shift plans. The crew scheduling plan uses the driver and crew rotation points, depots, and parking lots on the subway line as the dividing points, and divides the train running line into crew segments for drivers and crew members to take on duty. The crew segments are then combined into several anonymous crew tasks based on relevant constraints. Each crew task represents a driver and crew member's work schedule for one day. The crew shift plan is a three-type crew task generated according to the crew daily plan: morning, day, and night. Within a certain compilation cycle, the crew tasks are relatively evenly distributed to each driver and crew member according to the rotation system to form a stage crew plan.
[0003] Crew scheduling is the premise for the preparation of crew shift plans, and for a long time, the plan has been prepared manually. Due to the large scale of the problem and the complex constraints, the crew scheduling plan often takes a long time to prepare, and there are phenomena such as unbalanced task volume and low efficiency of drivers and crew members. With the extension of urban rail lines and the continuous increase in passenger demand, the manual preparation method has been unable to meet the actual operation needs, greatly limiting the operation efficiency, service quality and economic benefits of urban rail transit; how to quickly prepare efficient crew plans and get rid of the constraints of manual preparation has become an urgent problem to be solved.
[0004] Therefore, a subway crew scheduling management system and method based on MTSP is needed to solve the above technical problems. Summary of the invention
[0005] In order to solve the above technical problems, the present invention provides a subway crew scheduling management method based on MTSP, which obtains the optimal crew scheduling plan through the following steps: Step 1: Determine the rotation stations: Determine the rotation stations, depots and parking lots in the subway line according to the actual operation situation; Step 2: Crew segment division: Using the rotation station, depot and parking lot as the segmentation points, the subway line is divided into n crew segments, and a subway crew scheduling model is constructed based on the MTSP problem; The subway crew scheduling model abstracts crew segments into nodes with spatiotemporal attributes, abstracts the continuation of crew segments into arcs between spatiotemporal nodes, and introduces virtual start and end points to represent the attendance and departure of drivers and crew members. Step 3: Crew task generation: Starting from the starting point, the driver and crew select crew operation segments that meet the continuation conditions, combine them to form various crew tasks under the constraint of not exceeding the maximum duration of a single crew task, and form a crew scheduling plan through various crew tasks; Among them, the crew segments are arranged in the order of the start time and divided according to the task type; in the crew task, the crew segments can only be connected in one direction, and the crew segments of different task types cannot be connected; Step 4: Improve crew tasks: Solve the MTSP problem through the simulated annealing algorithm, find the optimal crew task combination covering all crew segments, and obtain the optimal crew scheduling plan.
[0006] As a further solution, the simulated annealing algorithm performs iterative optimization through the following steps: S101: Initialize simulated annealing parameters, including the maximum number of iterations of the population , crossover probability , mutation probability , initial simulated annealing temperature , Cooling rate ,parameter and the maximum number of annealing iterations ; S102: Initialize the population; construct the initial population by generating the initial solution to obtain the initial crew scheduling plan; S103: Calculate the population fitness of the initial population; solve the fitness function value of each individual in the population and the objective function value of the entire crew scheduling plan; S104: Determine whether the maximum number of iterations of the population has been reached ; If the maximum number of iterations of the population is reached , then jump to execute S113 operation; otherwise, execute S105 operation; S105: performing a selection operation; wherein the population is screened based on the fitness function value of the individual, and the parent generation individuals are selected to enter the child generation; S106: Perform a crossover operation; wherein, based on the crossover probability Perform crossover operation on offspring individuals; S107: Perform mutation operation; wherein, based on the mutation probability Perform mutation operations on offspring individuals; S108: Generate new population And used as the initial solution for simulated annealing; S109: Perturb the initial solution according to the set perturbation strategy to generate a new solution and as a candidate solution; S110: Set the parameters Used in Metropolis function module; determine the acceptance probability of candidate solutions and decide whether to accept the candidate solutions; S111: Cooling annealing; according to the set cooling rate Cool down and update the simulated annealing temperature value T ; S112: Determine whether the maximum number of annealing iterations has been reached ; If not, return to execute S109 operation; if reached, return to execute S103 operation; S113: Output the optimal solution of the population and map it to the optimal crew task combination to obtain the optimal crew scheduling plan.
[0007] As a further solution, the parameter When used in the Metropolis function module, the acceptance probability is calculated using the following formula: ; in, represents the initial solution, represents a candidate solution, represents the acceptance probability of the new solution, represents the simulated annealing temperature, Represent the fitness function values of the initial solution and candidate solution respectively.
[0008] As a further solution, the fitness function is set to the inverse of the objective function, which is a dual optimization objective function, including the shortest crew segment connection time objective function. and the workload balance objective function of drivers and passengers ;in, In the formula, and Respectively represent different crew segments, Indicates crew segment and The connection time between For the crew segment, represents the crew task set, j represents the crew task number, represents the penalty factor. Indicates the driver and passenger's convenient time. Crew clip Whether to continue the segment , Indicates crew tasks Is it in the optimal solution? Indicates Class cabin crew tasks, represents the total number of drivers and passengers, i represents the driver and passenger number, represents the working time of driver i, It represents the average working time of drivers and passengers; represents the objective function, Represents population The fitness function value of The objective function for minimizing the duration of crew segment connection The weight value of is the workload balance objective function for drivers and passengers The weight value of .
[0009] As a further solution, the initial solution generation is performed through the following steps: S201: Initialize the parameters for generating the population; including generating the initial population size And the crew segment and lower limit ; S202: Select crew segments that meet the connection time constraint condition and randomly divide all crew segments into Groups are assigned to each individual in the population to form corresponding crew tasks; S203: Check the feasibility of the crew task; traverse each individual in the population and check whether each generated crew task satisfies the value multiplication constraint condition; if not, return to execute step S202; if both are satisfied, execute step S204; S204: Determine whether all crew tasks meet feasibility requirements; if there are infeasible crew tasks, return to step S202 to regenerate the population; if feasibility requirements are met, output the initialized population.
[0010] As a further solution, the duty constraints include maximum on-duty time constraint, continuous duty time constraint, break time constraint, meal time constraint, space connection constraint, crew segment coverage constraint and shift working time constraint.
[0011] As a further solution, during the selection operation, an exponential transformation is performed on the basis of the selection roulette to determine the probability of an individual being selected; wherein, when the selection probability is subjected to exponential operation, the parameters are adjusted Controls the magnitude of the exponential transformation; In the formula, For individuals i The probability of being selected, the control parameter The value of adopts an adaptive mechanism and gradually decreases with the number of population iterations; Represents an individual i Fitness function value.
[0012] As a further solution, the perturbation strategy is as follows: randomly select a crew task from the current crew schedule for decomposition, and recombine the crew fragments obtained after the decomposition with the remaining crew tasks; wherein, when recombining: each crew fragment after decomposition is connected to the corresponding crew task; the combined crew task satisfies all constraints before the reorganization.
[0013] A subway crew scheduling management system based on MTSP, applied to a subway crew scheduling management method based on MTSP as described in any one of the above items, comprising an operation plan management unit, a model parameter management unit, a calculation function management unit, a constraint condition monitoring unit, a simulated annealing execution unit and an iterative solution management unit; wherein, Operation plan management unit: used for subway crew scheduling and crew task improvement management; including rotation station management module, crew segment management module, crew task management module and task improvement management module; Model parameter management unit: used to manage various parameters used in the subway crew scheduling model; including set parameter management module, model parameter management module and variable parameter management module; Calculation function management unit: used to manage the calculation function used by the simulated annealing algorithm; including the objective function management module and the fitness function management module; Constraint monitoring unit: used to monitor whether the crew scheduling plan meets the constraints; including the continuation constraint monitoring module and the duty constraint monitoring module; Simulated annealing execution unit: used to execute the simulated annealing algorithm and solve the MTSP problem; including the population update management module and the cooling annealing management module; Iterative solution management unit: used to manage the simulated annealing algorithm and obtain the optimal crew scheduling plan; including initialization parameter module, initialization population module, population iteration management module and annealing iteration management module.
[0014] As a further solution, it also includes a scheduling plan visualization analysis unit; which includes a subway line visualization module, a crew segment duration visualization module, a crew scheduling visualization module, a crew task working duration visualization module, and a crew task connection time visualization module.
[0015] Compared with the related art, the subway crew scheduling management system and method based on MTSP provided by the present invention has the following beneficial effects: The present invention transforms the subway crew scheduling plan into a multiple traveling salesman problem (MTSP). Based on a hierarchical sequence strategy, the longest on-duty time constraint, continuous duty time constraint, break time constraint, meal time and other constraints of each shift are comprehensively considered. The shortest crew segment connection time and balanced crew workload are taken as optimization goals to establish a subway crew scheduling model. Based on the MTSP solution idea, a genetic simulated annealing hybrid algorithm is designed to solve the model. The optimization effect in terms of the number of crew tasks and the length of the crew segment connection time is more significant, which can effectively reduce the labor cost of the enterprise and improve the duty efficiency of drivers and crew members. The problem caused by the manual preparation of crew scheduling plans in the prior art is solved. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 A schematic diagram of a subway crew scheduling management method based on MTSP provided by the present invention; Figure 2 A schematic diagram of a crew segment and crew tasks provided by the present invention; Figure 3 A schematic diagram for generating a crew task provided by the present invention; Figure 4 A schematic diagram of the hierarchical sequence strategy provided by the present invention; Figure 5 A schematic diagram of crew segment coding provided by the present invention; Figure 6 A schematic diagram of single-point and multi-point crossover provided by the present invention; Figure 7 A schematic diagram of the variation operation provided by the present invention; Figure 8 A schematic diagram of the disturbance operation provided by the present invention; Fig. 9 A flow chart of the genetic simulated annealing algorithm provided by the present invention; Fig.10 A schematic diagram of a subway line provided by the present invention; Fig.11 A distribution diagram of the duration of the crew segments provided by the present invention; Fig.12 A visualization diagram of the crew scheduling scheme provided by the present invention; Fig.13 The distribution diagram of the duration of the morning, day and night shift crew tasks provided by the present invention Fig.14 A schematic diagram of the crew task connection time provided by the present invention. DETAILED DESCRIPTION
[0017] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.
[0018] Embodiment 1 See also Figure 1 This embodiment provides a subway crew scheduling management method based on MTSP, which obtains the optimal crew scheduling plan through the following steps: Step 1: Determine the rotation stations: Determine the rotation stations, depots and parking lots in the subway line according to the actual operation situation; Step 2: Crew segment division: Using the rotation station, depot and parking lot as the segmentation points, the subway line is divided into n crew segments, and a subway crew scheduling model is constructed based on the MTSP problem; The subway crew scheduling model abstracts crew segments into nodes with spatiotemporal attributes, abstracts the continuation of crew segments into arcs between spatiotemporal nodes, and introduces virtual start and end points to represent the attendance and departure of drivers and crew members. Step 3: Crew task generation: Starting from the starting point, the driver and crew select crew operation segments that meet the continuation conditions, combine them to form various crew tasks under the constraint of not exceeding the maximum duration of a single crew task, and form a crew scheduling plan through various crew tasks; Among them, the crew segments are arranged in the order of the start time and divided according to the task type; in the crew task, the crew segments can only be connected in one direction, and the crew segments of different task types cannot be connected; Step 4: Improve crew tasks: Solve the MTSP problem through the simulated annealing algorithm, find the optimal crew task combination covering all crew segments, and obtain the optimal crew scheduling plan.
[0019] It should be noted that the crew scheduling plan is based on the train operation plan and the vehicle bottom utilization plan. The train operation line is divided into crew segments according to the line rotation station, vehicle depot and parking lot, and the crew work tasks are generated by considering the relevant value constraints of the driver and crew. Figure 2 As shown in the figure, the train running line on this line is divided into crew segments according to the rotation station. The crew segments that meet the connection time requirements and location constraints can be combined to form crew tasks. When all segments are covered by crew tasks, the crew scheduling plan is obtained. In order to reduce the operating costs of the enterprise and realize the rational allocation of resources, the crew scheduling plan should improve the duty efficiency of drivers and crew members as much as possible, reduce the number of drivers and crew members, and also consider the balance of the workload of drivers and crew members.
[0020] The Traveling Salesman Problem (TSP) is a classic combinatorial optimization problem. Suppose a traveling salesman starts from a starting city, passes through several other cities, and finally returns to the starting city. The total length of the path taken by the traveling salesman is required to be the shortest. The Multiple Traveling Salesman Problem (MTSP) is an extension of the Single Traveling Salesman Problem. Compared with the standard TSP, the MTSP requires that each city can only be visited by one traveling salesman, and the total path length of all traveling salesmen must be minimized.
[0021] Based on the description of crew scheduling and MTSP problem, both can be reduced to the optimal path planning problem in essence. Therefore, the crew scheduling can be converted into the MTSP problem for modeling and solving. Analogous to the setting in the MTSP problem, the crew scheduling plan to find the optimal crew tasks for all drivers and passengers can be compared to the MTSP problem to find the shortest walking path for all traveling salesmen; the crew segments in the crew scheduling plan can be compared to the city nodes in the MTSP problem, and the connection time of the crew segments is analogous to the walking distance between cities in the MTSP problem.
[0022] Therefore, a subway crew scheduling model is constructed based on the MTSP problem idea. The crew segments are abstracted as nodes with spatiotemporal attributes, and the connection between crew operation segments is abstracted as the connection arcs between spatiotemporal nodes. At the same time, virtual starting points and end points are introduced as crew supply centers to represent the attendance and off-duty of drivers and crew members. Different from the MTSP problem, the connection of crew segments in the crew scheduling plan network diagram needs to consider complex spatiotemporal constraints, and not any crew segments can be connected.
[0023] Figure 3 To generate a schematic diagram for a crew task, the driver and crew start from the starting point, select crew operation segments that meet the connection conditions, and combine them to form each crew task under the constraint of not exceeding the maximum duration of a single crew task. The sequence numbers of the crew segments are arranged in the order of the start time of the crew segments, and are divided according to the task type. Therefore, the crew segments can only be connected in one direction and different types of crew segments cannot be connected.
[0024] Therefore, this embodiment decomposes the scheduling problem into tasks at different levels and establishes associations between these levels to achieve a more efficient scheduling solution. Based on the description of the subway crew scheduling problem, the subway crew scheduling can be divided into the following four levels. The hierarchical sequence strategy is as follows: Figure 4 shown.
[0025] (1) Level 1: Determination of the transfer station This level needs to determine the rotation stations for drivers and crew members in the subway line according to actual operational needs. The line rotation stations, vehicle depots, and parking lots are all key nodes in the subway crew scheduling plan, which directly affect the results of the crew segments.
[0026] (2) Level 2: Crew segment division This level is based on the train operation diagram and the vehicle operation plan, and divides the train line into service segments with the wheel station, vehicle depot and parking lot as the dividing points. Assume that a subway line is divided into the train line through the wheel point. A crew segment, For crew fragment collection, Represents the set of vehicle bottom numbers of the crew segment. A single crew segment is shown in formula (1). Indicates the number of the crew segment, Indicates the vehicle number of the crew segment, Indicates the start time of the crew segment, Indicates the end time of the crew segment, Indicates the start point of the crew segment, Indicates the end point of the crew segment.
[0027] (1) (3) Level 3: Crew Task Generation This level needs to consider the constraints related to the driver and crew's duty and combine the crew segments into crew tasks. A crew task generally includes 3-5 crew segments, and the total duration does not exceed the maximum working time limit of the driver and crew.
[0028] A single crew task represents the entire process from the driver's attendance to the departure. Indicates Class crew task set ( , respectively morning shift, day shift, and night shift), use Indicates The number of the class crew task set is The crew task ( ), as shown in formula (2), where The number of crew segments included in the crew task.
[0029] (2) (4) Level 4: Crew Task Improvement This level needs to optimize and improve the crew tasks generated by the combination according to certain optimization goals. Common optimization and improvement methods include crew task decomposition and combination, crew segment exchange, etc. The optimization goals include the balance of crew working time, the minimum number of crew tasks, and the highest duty efficiency of drivers and crew members. While improving the crew tasks, it is necessary to ensure that the newly generated crew tasks always meet the relevant duty constraints of drivers and crew members. The relevant constraints include the longest on-duty time, the longest continuous driving time, the break and meal time, and the space connection constraints. Finally, a crew scheduling plan covering all crew segments and with good optimization effect on the target is generated; the symbols and meanings of the relevant parameter variables are shown in Tables 1-3.
[0030] Table 1 Collection parameter explanation Table 2 Explanation of model parameters Table 3 Explanation of variable parameters like Fig. 9 As shown in Figure 2, as a further solution, the simulated annealing algorithm performs iterative optimization through the following steps: S101: Initialize simulated annealing parameters, including the maximum number of iterations of the population , crossover probability , mutation probability , initial simulated annealing temperature , Cooling rate ,parameter and the maximum number of annealing iterations ; S102: Initialize the population; construct the initial population by generating the initial solution to obtain the initial crew scheduling plan; S103: Calculate the population fitness of the initial population; solve the fitness function value of each individual in the population and the objective function value of the entire crew scheduling plan; S104: Determine whether the maximum number of iterations of the population has been reached ; If the maximum number of iterations of the population is reached , then jump to execute S113 operation; otherwise, execute S105 operation; S105: performing a selection operation; wherein the population is screened based on the fitness function value of the individual, and the parent generation individuals are selected to enter the child generation; S106: Perform a crossover operation; wherein, based on the crossover probability Perform crossover operation on offspring individuals; S107: Perform mutation operation; wherein, based on the mutation probability Perform mutation operations on offspring individuals; S108: Generate new population And used as the initial solution for simulated annealing; S109: Perturb the initial solution according to the set perturbation strategy to generate a new solution and as a candidate solution; S110: Set the parameters Used in Metropolis function module; determine the acceptance probability of candidate solutions and decide whether to accept the candidate solutions; S111: Cooling annealing; according to the set cooling rate Cool down and update the simulated annealing temperature value T ; S112: Determine whether the maximum number of annealing iterations has been reached ; If not, return to execute S109 operation; if reached, return to execute S103 operation; S113: Output the optimal solution of the population and map it to the optimal crew task combination to obtain the optimal crew scheduling plan.
[0031] It should be noted that commonly used intelligent optimization algorithms include genetic algorithm, ant colony algorithm, simulated annealing algorithm, taboo search algorithm, etc. This embodiment aims at solving the MTSP problem and selects a genetic algorithm mixed with a simulated annealing algorithm for solving the problem in combination with the characteristics of the subway crew scheduling model.
[0032] In order to solve the subway crew scheduling problem, this paper designs a population in which each individual has a chromosome, corresponding to a crew task. Each gene on the chromosome represents a crew segment number, indicating the crew segment that the crew task needs to multiply. The specific encoding operation is shown as follows: Figure 5 shown.
[0033] (1) Selection This embodiment selects roulette and performs exponential transformation to determine the probability of an individual being selected, as shown in formula (17). Performing exponential operation on the selection probability can amplify or reduce the difference between fitness values. is a control parameter that controls the amplitude of the exponential transformation. When , the difference between fitness values will be magnified. The larger the value, the more obvious the effect of amplifying fitness differences. The value is between 0.1 and 1. The value of adopts an adaptive mechanism, which gradually decreases with the number of iterations. The value range is .
[0034] (2) Cross Based on the characteristics of subway crew scheduling, this paper adopts a single-point crossover ( Figure 6 Left) and the two-point crossover operation ( Figure 6 Right), the crossover probability is set using an adaptive mechanism. As the number of iterations increases, the crossover probability gradually decreases. Referring to relevant literature and the actual debugging process of the algorithm, the range of the crossover probability is set to .
[0035] (3) Variation Generally speaking, the mutation probability is usually set between 0.01 and 0.1. Considering the population size characteristics of the crew scheduling problem, setting a smaller mutation probability may not ensure that the mutation operation enriches the diversity of the population. Therefore, this paper sets the mutation probability For the mutation operation, the mutation gene position of the chromosome is randomly generated. Due to the particularity of the crew scheduling problem, the relevant crew segment continuation constraints must be met. If not, it is necessary to mutate again until the crew task is feasible. The mutation operation is as follows Figure 7 shown.
[0036] In the process of improving the genetic algorithm, a new population is generated through selection, crossover, and mutation operations, which is used as the initial solution of the simulated annealing algorithm. Based on the initial solution of the algorithm, a neighborhood candidate solution generation scheme (perturbation strategy) is designed to achieve a more refined search in the neighborhood.
[0037] The initial parameters of the simulated annealing algorithm include initial temperature, cooling rate, number of iterations, etc. In the simulated annealing algorithm, the setting of the initial temperature has a significant impact on the algorithm's search ability and convergence speed; the cooling rate affects the speed at which the temperature drops. If the cooling rate is too small, the algorithm tends to converge quickly and fall into a local optimal solution; on the contrary, if the cooling rate is too low, the algorithm's global search ability increases, and the algorithm's convergence speed will also decrease; in addition, a reasonable number of iterations needs to be set to take into account both the algorithm's search ability and solution efficiency. According to multiple debugging tests, the initial parameter settings are shown in Table 4.
[0038] Table 4 Initial parameter setting table of simulated annealing The perturbation strategy refers to generating candidate solutions in the neighborhood of the current solution. In the design of the perturbation strategy in solving the TSP problem, combined with the characteristics of the subway crew scheduling problem, this paper designs a "decomposition-combination" perturbation strategy. The basic idea is to randomly select a crew task from the current scheduling plan to decompose it, and then recombine the decomposed crew fragments with several crew tasks.
[0039] The combination of crew tasks should meet two conditions: ensure that all crew segments are connected to specific crew tasks after decomposition; ensure that the combined crew tasks meet the relevant driver and crew duty constraints. Figure 8 shown.
[0040] Frequent perturbation operations may cause the algorithm to fall into a local optimal solution. It is necessary to set an adaptive mechanism to control the frequency of perturbations. The perturbation frequency function is set as shown in formula (18).
[0041] (18) in, It is The perturbation frequency at the iteration, is the initial perturbation frequency, which is usually set larger to ensure a wide search in the initial stage. is the termination perturbation frequency, is the number of iterations. After many experimental attempts, Set to 0.85, Set to 0.13.
[0042] As a further solution, the parameter When used in the Metropolis function module, the acceptance probability is calculated using the following formula: ; in, represents the initial solution, represents a candidate solution, represents the acceptance probability of the new solution, represents the simulated annealing temperature, Represent the fitness function values of the initial solution and candidate solution respectively.
[0043] It should be noted that in the simulated annealing algorithm, the Metropolis criterion is used to determine whether to accept a candidate solution. The probability of accepting a candidate solution is determined by the difference in the objective function value between the optimal solution and the candidate solution and the temperature value during the iteration process. Improved Metropolis criterion, this parameter is used to flexibly adjust the weights of the two factors of target value difference and temperature value in the probability calculation of candidate solutions; if the parameter If the value is too small, the probability of accepting the candidate solution will be more sensitive to the difference in the objective function value, which can amplify the impact of the difference in the objective function value on the probability; otherwise, the probability of accepting the candidate solution will be mainly affected by the annealing temperature. Set the value of to 2.
[0044] As a further solution, the fitness function is set to the inverse of the objective function, which is a dual optimization objective function, including the shortest crew segment connection time objective function. and the workload balance objective function of drivers and passengers ;in, In the formula, and Respectively represent different crew segments, Indicates crew segment and The connection time between For the crew segment, represents the crew task set, j represents the crew task number, represents the penalty factor. Indicates the driver and passenger's convenient time. Crew clip Whether to continue the segment , Indicates crew tasks Is it in the optimal solution? Indicates Class cabin crew tasks, represents the total number of drivers and passengers, i represents the driver and passenger number, represents the working time of driver i, It represents the average working time of drivers and passengers; represents the objective function, Represents population The fitness function value of The objective function for minimizing the duration of crew segment connection The weight value of is the workload balance objective function for drivers and passengers The weight value of .
[0045] It should be noted that the preparation of subway crew scheduling is closely related to the company's operating costs, the work efficiency and salary of drivers and crew members. This article comprehensively considers the interests of both parties and sets the optimization goals of the subway crew scheduling plan as the shortest connection time of crew segments and balanced workload of drivers and crew members.
[0046] (1) The crew segment has the shortest duration There are two cases for optimizing the continuation time of crew segments. One is normal continuation, where the end point of the crew segment being continuation is the same as the start point of the crew segment being continuation, that is, ; The second is the convenient ride connection. The end point of the service segment being connected is inconsistent with the start point of the connecting service segment. The driver and crew need to take the train to the destination to perform the new service segment. .
[0047] When the crew segment is considered to be connected normally, the objective function is shown in formula (11). Representative crew clip and The connection time between them, when the connection time satisfies the interval time constraint, is , otherwise it is set to infinite M, indicating that the crew segment cannot be continued.
[0048] (11) The impact of the on-duty efficiency of drivers and crew members is relatively large, so the occurrence of on-duty situations should be reduced as much as possible when compiling crew tasks. Therefore, a penalty factor is introduced for on-duty continuity. Construct the objective function of the multiplication continuation, see formula (12).
[0049] (12) Combining the objective functions of the two connection situations, we can obtain the optimization goal of minimizing the total connection time of the crew segment, as shown in formula (13): (13) (2) Balanced workload for drivers and passengers Considering that the salary of subway drivers and passengers in my country is calculated based on "driving kilometers", in order to mobilize the work enthusiasm of drivers and passengers and ensure the balance of their salaries, it is possible to consider balancing the workload of drivers and passengers as the optimization goal in the crew scheduling plan. In order to reflect the balance of the workload of drivers and passengers, the optimization goal can be set to minimize the variance of the working time of drivers and passengers. The working time of all drivers and passengers is defined as , is the average working time of all drivers and passengers, then the objective function of driver and passenger workload balance is as shown in formula (14).
[0050] (14) In summary, this paper constructs two optimization objective functions, selecting the shortest crew segment connection time as the main optimization object and the balance of driver and passenger workload as the secondary optimization objective. The problem is transformed into a single-objective optimization model, as shown in formula (15): (15) As a further solution, the initial solution generation is performed through the following steps: S201: Initialize the parameters for generating the population; including generating the initial population size And the crew segment and lower limit ; S202: Select crew segments that meet the connection time constraint condition and randomly divide all crew segments into Groups are assigned to each individual in the population to form corresponding crew tasks; S203: Check the feasibility of the crew task; traverse each individual in the population and check whether each generated crew task satisfies the value multiplication constraint condition; if not, return to execute step S202; if both are satisfied, execute step S204; S204: Determine whether all crew tasks meet feasibility requirements; if there are infeasible crew tasks, return to step S202 to regenerate the population; if feasibility requirements are met, output the initialized population.
[0051] It should be noted that the initial population is the starting point of the iterative optimization of the genetic algorithm. This embodiment considers the multiplication constraints of the relevant values of the drivers and passengers, and uses a random generation method to construct the initial population; the crew task set corresponding to all individuals in the initial population is the initial crew scheduling plan.
[0052] As a further solution, the duty constraints include maximum on-duty time constraint, continuous duty time constraint, break time constraint, meal time constraint, space connection constraint, crew segment coverage constraint and shift working time constraint.
[0053] (1) Maximum on-duty time constraints The maximum on-duty time refers to the longest time interval between the subway drivers and crew members going to work and leaving get off work. It stipulates the upper limit of the drivers and crew members' working hours in a day. There are also differences in the maximum on-duty time requirements for the three different types of crew tasks: morning shift, day shift and evening shift.
[0054] (3) (2) Continuous value multiplication of time constraints Continuous duty time refers to the time that the driver or crew member completes a service segment and continues to drive the train to complete the next service segment without changing shifts. Continuous duty tasks usually consist of more than two service segments. In the process of generating hierarchical sequence service tasks, it is necessary to determine whether the continuation of the service segments meets the continuous duty time constraint.
[0055] (4) (3) Rest time constraints To avoid fatigue driving, drivers and passengers need to take a rest after completing certain driving tasks, and the rest time must meet the maximum and minimum rest time constraints. Rest time is an important basis for constructing the crew segment connection time matrix. If the interval between two crew segments is within the maximum and minimum rest time and meets the spatial connection constraint, the driver and passengers can be arranged to take a rest, and the rest time is the connection time of the crew segments.
[0056] (5) (4) Meal time restrictions The start and end time of the morning shift crew task is relatively early, so generally no breakfast is arranged for the drivers and crew, and only lunch and dinner are arranged. If the end time of a certain crew segment of the driver and crew falls within the meal time window, the meal is arranged, and the meal time should be controlled within the minimum meal time and the maximum meal time range.
[0057] (6) (7) (5) Spatial connection constraints The spatial connection constraint requires that the terminal station of the previous crew segment in the adjacent crew segments that the driver and crew are on duty must be the same as the starting station of the next crew segment. When the spatial connection condition is not met, the driver and crew need to be arranged to take other trains as passengers to reach the starting station of the next crew segment. From the perspective of optimization, the occurrence of taking other trains should be avoided as much as possible.
[0058] (8) (6) Crew segment coverage constraints In the subway crew scheduling plan, all crew segments must have corresponding drivers on duty, that is, the crew task set must cover all crew segments.
[0059] (9) (7) Shift working time constraints.
[0060] Each shift type requires that all the attendance and departure times of this type of shifts are within the shift time requirements. Tsδ represents the start time of this type of shift, and Teδ represents the end time of this type of shift. The time requirements of the three types of shifts are different. The earliest start time of the morning shift is the start of operation, and the latest end time is set at 11:00; the earliest start time of the day shift is set at 9:00, and the latest end time is set at 18:00; the earliest start time of the night shift is 17:00, and the latest end time is set at the end of operation.
[0061] (10) As a further solution, during the selection operation, an exponential transformation is performed on the basis of the selection roulette to determine the probability of an individual being selected; wherein, when the selection probability is subjected to exponential operation, the parameters are adjusted Controls the magnitude of the exponential transformation; (17) In the formula, For individuals i The probability of being selected, the control parameter (The value of adopts an adaptive mechanism and gradually decreases with the number of population iterations; Represents an individual i Fitness function value.
[0062] As a further solution, the perturbation strategy is as follows: randomly select a crew task from the current crew schedule for decomposition, and recombine the crew fragments obtained after the decomposition with the remaining crew tasks; wherein, when recombining: each crew fragment after decomposition is connected to the corresponding crew task; the combined crew task satisfies all constraints before the reorganization.
[0063] Embodiment 2 A subway crew scheduling management system based on MTSP, applied to a subway crew scheduling management method based on MTSP as described in any one of the above items, comprises an operation plan management unit, a model parameter management unit, a calculation function management unit, a constraint condition monitoring unit, a simulated annealing execution unit and an iterative solution management unit; wherein, Operation plan management unit: used for subway crew scheduling and crew task improvement management; including rotation station management module, crew segment management module, crew task management module and task improvement management module; Model parameter management unit: used to manage various parameters used in the subway crew scheduling model; including set parameter management module, model parameter management module and variable parameter management module; Calculation function management unit: used to manage the calculation function used by the simulated annealing algorithm; including the objective function management module and the fitness function management module; Constraint monitoring unit: used to monitor whether the crew scheduling plan meets the constraints; including the continuation constraint monitoring module and the duty constraint monitoring module; Simulated annealing execution unit: used to execute the simulated annealing algorithm and solve the MTSP problem; including the population update management module and the cooling annealing management module; Iterative solution management unit: used to manage the simulated annealing algorithm and obtain the optimal crew scheduling plan; including initialization parameter module, initialization population module, population iteration management module and annealing iteration management module.
[0064] As a further solution, it also includes a scheduling plan visualization analysis unit; which includes a subway line visualization module, a crew segment duration visualization module, a crew scheduling visualization module, a crew task working duration visualization module, and a crew task connection time visualization module.
[0065] It should be noted that: in a specific embodiment, the subway line visualization module displays Fig.10For the subway line shown in the figure, the train operation line is divided into crew operation segments based on the rotation station, and a total of 686 crew operation segments are obtained. The crew operation segments are arranged in the order of the starting time, as shown in Table 5.
[0066] Table 5 Flight attendant operation fragment table The crew segment duration visualization module is used to visualize the operation time of 686 crew segments, such as Fig.11 As shown in the figure, more than 95% of the crew segments’ operation time is concentrated between 20 minutes and 30 minutes and between 70 minutes and 80 minutes, and the duration distribution of crew segments is relatively balanced.
[0067] Before solving the crew scheduling model, the parameters used need to be set, including the maximum on-duty time, continuous duty time, shortest rest time, longest rest time, meal time and meal time window, etc. The following sets the relevant parameters according to the specific conditions of the subway line and relevant rules and regulations. The parameter time standard values are shown in Table 6 below.
[0068] Table 6 Time parameter value table The algorithm finally obtained 110 crew tasks, including 37 morning shift crew tasks, 43 day shift crew tasks, and 30 night shift crew tasks. The corresponding relationship between the crew tasks and crew segments in the final subway crew scheduling plan is presented through the crew scheduling visualization module, as shown below Fig.12 Among them, red, green and blue represent the morning shift crew segment, day shift crew segment and night shift crew segment respectively.
[0069] Tables 7 to 9 describe some of the crew tasks for the morning, daytime, and night shifts, respectively. It should be noted that due to the disorder of the algorithm during the search process, the numbers of the crew tasks are not arranged in sequence.
[0070] Table 7 Morning shift scheduling plan Table 8 Day shift scheduling plan Table 9 Night shift scheduling plan Different crew tasks have different working hours for drivers and crew members. The crew task working hours visualization module counts the crew task hours of the morning shift, day shift, and night shift respectively and displays them in a line graph. Fig.13It can be seen that the duration of morning shift crew tasks is mostly distributed between 400~480 minutes and 50~150 minutes, the duration of day shift crew tasks is mostly distributed between 400~480 minutes, and the duration of night shift crew tasks is more scattered. The difference in the working time of crew tasks is mainly related to the distribution of train lines in each time period.
[0071] The distribution of crew segment connection time in 110 crew tasks in the subway crew scheduling plan is presented through the crew task connection time visualization module (such as Fig.14 As shown). Under normal circumstances, drivers and crew members need to arrange 3 to 4 rest times and 1 meal time during a day's duty, so a connection time of about 100 minutes is more reasonable. It can be seen from the figure that more than 60% of the crew task connection time is within the range of 70 to 140 minutes, and the crew task connection is relatively reasonable; in addition, about 10% of the crew task connection time is less than 50 minutes, and more than 20% of the crew task connection time is about 200 minutes. Too short connection time leads to insufficient rest time for drivers and crew members, affecting the safe driving of trains. At the same time, too long connection time leads to too long waiting time for drivers and crew members, reducing the duty efficiency of drivers and crew members.
[0072] The above are only embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A subway crew scheduling management method based on MTSP, It is characterized in that The optimal crew scheduling plan is obtained through the following steps: Step 1: Determine the rotation stations: Determine the rotation stations, depots and parking lots in the subway line according to the actual operation situation; Step 2: Crew segment division: Using the rotation station, depot and parking lot as the segmentation points, the subway line is divided into n crew segments, and a subway crew scheduling model is constructed based on the MTSP problem; The subway crew scheduling model abstracts crew segments into nodes with spatiotemporal attributes, abstracts the continuation of crew segments into arcs between spatiotemporal nodes, and introduces virtual start and end points to represent the attendance and departure of drivers and crew members. Step 3: Crew task generation: Starting from the starting point, the driver and crew select crew operation segments that meet the continuation conditions, combine them to form various crew tasks under the constraint of not exceeding the maximum duration of a single crew task, and form a crew scheduling plan through various crew tasks; Among them, the crew segments are arranged in the order of the start time and divided according to the task type; in the crew task, the crew segments can only be connected in one direction, and the crew segments of different task types cannot be connected; Step 4: Improve crew tasks: Solve the MTSP problem through the simulated annealing algorithm, find the optimal crew task combination covering all crew segments, and obtain the optimal crew scheduling plan.
2. According to the MTSP-based subway crew scheduling management method of claim 1, It is characterized in that The simulated annealing algorithm performs iterative optimization through the following steps: S101: Initialize simulated annealing parameters, including the maximum number of iterations of the population , crossover probability , mutation probability , initial simulated annealing temperature , Cooling rate ,parameter and the maximum number of annealing iterations ; S102: Initialize the population; construct the initial population by generating the initial solution to obtain the initial crew scheduling plan; S103: Calculate the population fitness of the initial population; solve the fitness function value of each individual in the population and the objective function value of the entire crew scheduling plan; S104: Determine whether the maximum number of iterations of the population has been reached ; If the maximum number of iterations of the population is reached , then jump to execute S113 operation; otherwise, execute S105 operation; S105: performing a selection operation; wherein the population is screened based on the fitness function value of the individual, and the parent generation individuals are selected to enter the child generation; S106: Perform a crossover operation; wherein, based on the crossover probability Perform crossover operation on offspring individuals; S107: Perform mutation operation; wherein, based on the mutation probability Perform mutation operations on offspring individuals; S108: Generate new population And used as the initial solution for simulated annealing; S109: Perturb the initial solution according to the set perturbation strategy to generate a new solution and as a candidate solution; S110: Set the parameters Used in Metropolis function module; determine the acceptance probability of candidate solutions and decide whether to accept the candidate solutions; S111: Cooling annealing; according to the set cooling rate Cool down and update the simulated annealing temperature value T ; S112: Determine whether the maximum number of annealing iterations has been reached ; If not, return to execute S109 operation; if reached, return to execute S103 operation; S113: Output the optimal solution of the population and map it to the optimal crew task combination to obtain the optimal crew scheduling plan.
3. According to the MTSP-based subway crew scheduling management method of claim 2, It is characterized in that The parameters When used in the Metropolis function module, the acceptance probability is calculated using the following formula: ; in, represents the initial solution, represents a candidate solution, represents the acceptance probability of the new solution, represents the simulated annealing temperature, Represent the fitness function values of the initial solution and candidate solution respectively.
4. According to the MTSP-based subway crew scheduling management method of claim 2, It is characterized in that The fitness function is set to the inverse of the objective function, which is a dual optimization objective function, including the shortest crew segment connection time objective function. and the workload balance objective function of drivers and passengers ;in, In the formula, and Respectively represent different crew segments, Indicates crew segment and The connection time between For the crew segment, represents the crew task set, j represents the crew task number, It means multiplying the penalty factor. Indicates the driver and passenger's convenient time. Crew clip Whether to continue the segment , Indicates crew tasks Is it in the optimal solution? Indicates Class cabin crew tasks, represents the total number of drivers and passengers, i represents the driver and passenger number, represents the working time of driver i, It represents the average working time of drivers and passengers; represents the objective function, Represents population The fitness function value of The objective function for minimizing the duration of crew segment connection The weight value of is the workload balance objective function for drivers and passengers The weight value of .
5. According to the MTSP-based subway crew scheduling management method of claim 2, It is characterized in that The initial solution is generated by the following steps: S201: Initialize the parameters for generating the population; including generating the initial population size And the crew segment and lower limit ; S202: Select crew segments that meet the connection time constraint condition and randomly divide all crew segments into Groups are assigned to each individual in the population to form corresponding crew tasks; S203: Check the feasibility of the crew task; traverse each individual in the population and check whether each generated crew task satisfies the value multiplication constraint condition; if not, return to execute step S202; if both are satisfied, execute step S204; S204: Determine whether all crew tasks meet the feasibility requirements; if there are infeasible crew tasks, return to step S202 to regenerate the population; if the feasibility requirements are met, output the initialized population.
6. According to the MTSP-based subway crew scheduling management method of claim 5, It is characterized in that The duty constraints include maximum on-duty time constraints, continuous duty time constraints, break time constraints, meal time constraints, space connection constraints, crew segment coverage constraints and shift working time constraints.
7. According to the MTSP-based subway crew scheduling management method of claim 2, It is characterized in that In the selection operation, an exponential transformation is performed on the basis of the selection roulette to determine the probability of an individual being selected; when the selection probability is exponentially calculated, the parameters are adjusted Controls the magnitude of exponential transformations; In the formula, For individuals i The probability of being selected, the control parameters The value of adopts an adaptive mechanism and gradually decreases with the number of population iterations; Represents an individual i Fitness function value.
8. According to the MTSP-based subway crew scheduling management method of claim 2, It is characterized in that Perturbation strategy: randomly select a crew task from the current crew scheduling plan for decomposition, and recombine the crew fragments obtained after decomposition with the remaining crew tasks; when recombining: each crew fragment after decomposition is connected to the corresponding crew task; the combined crew task satisfies all constraints before reorganization.
9. A subway crew scheduling management system based on MTSP, applied to a subway crew scheduling management method based on MTSP as claimed in any one of claims 1 to 8, It is characterized in that It includes an operation plan management unit, a model parameter management unit, a calculation function management unit, a constraint monitoring unit, a simulated annealing execution unit and an iterative solution management unit; among which, Operation plan management unit: used for subway crew scheduling and crew task improvement management; including rotation station management module, crew segment management module, crew task management module and task improvement management module; Model parameter management unit: used to manage various parameters used in the subway crew scheduling model; including set parameter management module, model parameter management module and variable parameter management module; Calculation function management unit: used to manage the calculation function used by the simulated annealing algorithm; including the objective function management module and the fitness function management module; Constraint monitoring unit: used to monitor whether the crew scheduling plan meets the constraints; including the continuation constraint monitoring module and the duty constraint monitoring module; Simulated annealing execution unit: used to execute the simulated annealing algorithm and solve the MTSP problem; including the population update management module and the cooling annealing management module; Iterative solution management unit: used to manage the simulated annealing algorithm and obtain the optimal crew scheduling plan; including initialization parameter module, initialization population module, population iteration management module and annealing iteration management module.
10. A subway crew scheduling management system based on MTSP according to claim 9, It is characterized in that It also includes a scheduling plan visualization analysis unit; which includes a subway line visualization module, a crew segment duration visualization module, a crew scheduling visualization module, a crew task working duration visualization module, and a crew task connection time visualization module.
Citation Information
Patent Citations
Vehicle ride-sharing path optimization method and system
CN111667086A
Subway crew scheduling plan compilation optimization method based on SPFA algorithm
CN113837438A
Subway crew plan optimization compilation method based on travel path
CN116307494A
Multi-class bus scheduling method considering rule constraint and tabu search
CN117132045A
Cushion scheduling compilation optimization algorithm
CN117635095A
