Flexible Production Line Scheduling Method for Multiple Vehicle Models in Batches

Abstract

A scheduling method for a multi - model batch flexible production line. First, the configuration parameters of the production line are retrieved. Considering the impacts of random sampling events, rework random events, and workshop shutdown random events, the evaluation parameters of the solution and the parameters of the inner - layer simulated annealing algorithm are determined. Then, taking the downstream vehicle model retrieval sequence as the input, the two major decision variables of the minimum batch size and vehicle model sorting are divided into a two - layer optimization problem. Among them: after the minimum batch size is determined in the outer layer, the inner - layer simulated annealing algorithm solves the sorting and scheduling problem with the minimum batch size constraint to obtain the optimal production sorting solution. The present invention can solve the production scheduling problem with minimum batch size constraints and random event disturbances, providing strong decision - making support for the batch flexible production scenario in the real environment.

Description

Technical Field

[0001] The present invention relates to a technology in the field of flexible workshop production and manufacturing, specifically a multi-model batch flexible production line scheduling method. Background Art

[0002] In an actual multi-model flexible production line, when mixed-line production is carried out, it is necessary to combine the large Block manufacturing mode to achieve flexible production. A Block is the quantity of continuously producing the same model, also known as the minimum batch size for batching. For example, when Block = 20, the production model can be switched after every 20 vehicles. In production, it is necessary to first formulate a unified production Block, and then decide the specific model produced by each Block. The existing multi-model mixed-line flexible production scheduling methods are mostly based on the situation of complete flexibility of the production line, that is, the actual constraints of the minimum batch size are not considered. In addition, the existing flexible production scheduling methods also lack the consideration of random events such as sampling inspection, repair, and shutdown on the production line, and it is difficult to adapt to the actual production scenario. Summary of the Invention

[0003] In view of the above deficiencies of the existing technology, the present invention proposes a multi-model batch flexible production line scheduling method, which can solve the production scheduling problem with minimum batch size constraints and random event disturbances, and provide strong decision-making support for the batch flexible production scenario in the real environment.

[0004] The present invention is realized through the following technical solutions:

[0005] The present invention relates to a multi-model batch flexible production line scheduling method. First, the configuration parameters of the production line are retrieved, and considering the influence of random events such as sampling inspection, repair, and workshop shutdown, the evaluation parameters of the solution and the parameters of the inner-layer simulated annealing algorithm are determined; then, taking the downstream vehicle model retrieval order as the input, the two decision variables of the minimum batch size and vehicle model sorting are divided into two-layer optimization problems, where: after determining the minimum batch size in the outer layer, the inner-layer simulated annealing algorithm solves the sorting scheduling problem with minimum batch size constraints to obtain the optimal production sorting solution.

[0006] The configuration parameters of the production line include but are not limited to the multi-model retrieval order of the downstream general assembly workshop for the vehicle body distribution center, etc.

[0007] The multi-model batch flexible production line means that the production line has S workstations. According to the production plan, with 1 minute as 1 beat, the bodywork of multiple vehicle models is successively processed through the processes of S workstations and enters the vehicle body distribution center for storage as inventory, where the number of continuously produced bodies of each vehicle model is called the minimum batch size B (Block) for batching.

[0008] The described body distribution center refers to: a body warehouse with reordering function, which can receive the processed bodies from the body shop, sort them according to different vehicle models, and provide the required vehicle models to the general assembly shop.

[0009] The described multi-vehicle model retrieval sequence refers to: at each beat, the general assembly shop retrieves one body of a certain vehicle model from the body distribution center for overall assembly, and this retrieval sequence is a sequence D of length N bodies that is known in advance.

[0010] The described production sorting solution refers to: a production plan sequence P of length N bodies, indicating the body types input into the body shop for processing at each beat. After being processed through S workstations, these bodies enter the body distribution center for stocktaking.

[0011] The described random sampling event refers to: when a body enters the quality control area at each beat, the vehicle is randomly sampled with a probability of p check for a duration of T check beats. When a body is sampled, the subsequent bodies are still in other processes and cannot be accelerated to replace it, meaning that there will be a beat when the body shop does not output a body. After the sampling is completed, in order to reduce vehicle model switching, it is not immediately inserted into the production line, but inserted into the nearest sequence of the same vehicle model. When inserting, the workstations before the sampling need to pause and wait. There is a probability of p breakdown of finding quality problems during the inspection, causing the entire workshop to stop work for maintenance.

[0012] The described rework random event refers to: there is a probability of p repair that a body will require rework during the painting process and will be sent back to the body shop to repair the defects. The reworked body directly enters the body distribution center.

[0013] The described workshop shutdown random event refers to: the workshop has a time ratio of p malfunction in the state of production line failure. Each maintenance generally requires T breakdown beats. Assuming that the probability of shutdown at each beat is equal to p repair , the average non-shutdown duration is From which the shutdown probability of each beat can be deduced The failures are divided into two types. When the body production line stops, no sampling and rework work is carried out, and no new bodies enter the body distribution center. When the general assembly production line stops, the general assembly shop does not retrieve goods from the body distribution center.

[0014] The described evaluation parameters of the solution refer to: the coefficient weights of the number of vehicle model switches S, the number of part transports T, the number of parts stored at the line side R, and the number of shortages in the body distribution center L, and the sum is used to obtain the evaluation index C of the sorting solution.

[0015] The optimization objective is as follows: Minimize the evaluation index of the solution Min.C = α·S + β·T + δ·R + σ·L, where: The number of vehicle model switches S refers to the number of times of switching vehicle models. The algorithm traverses the production plan sequence from the beginning. Whenever the vehicle models produced by two consecutive batches are different, the number of switches is incremented by 1; The number of part transports T refers to the number of part transports. For each part, whenever a vehicle model is about to be produced and the corresponding part at the line side is insufficient, it is determined whether to transport one or two boxes of parts based on the next usage amount of this part. Each box of parts corresponds to only one vehicle model, and at most two boxes of parts can be stored at the line side. Therefore, the existing parts at the line side are transported back according to the next usage situation. Whenever a part is transported or transported back, the number of part transports is incremented by 1; The quantity of parts stored at the line side R refers to checking the number of boxes stored at the line side (0 or 1 or 2) at each beat. The quantity stored at the line side is the sum of these box numbers; The number of shortages at the body distribution center L refers to the number of shortages in the production process when adjusting the goods. When the general assembly workshop requests to adjust the goods from the body distribution center at each beat, if the body inventory of the corresponding vehicle model is 0, a shortage is recorded once.

[0016] The outer traversal described above includes roughly determining the minimum batch range and precise search:

[0017] In the first step, roughly determine the minimum batch. First, determine the possible range of positive integer values of the minimum batch according to the actual production: (b1, b2). Within this range, first use a smaller number of simulation times M = M1, traverse the specified minimum batch b1 ≤ B ≤ b2, input it into the inner simulated annealing algorithm, and obtain the corresponding optimal production sorting P of the inner layer B , so as to determine the optimal approximate range of the minimum batch B

[0018] In the second step, for precise search, within the range of (b3, b4), use a larger number of simulation times M = M2 > M1, traverse the specified minimum batch b3 ≤ B ≤ b4, input it into the inner simulated annealing algorithm, and obtain the corresponding optimal production sorting P of the inner layer B , and screen out the optimal evaluation value C(P B ) corresponding to the minimum batch B and production sorting P B , which is output as the optimal solution.

[0019] The inner simulated annealing algorithm, based on the general assembly workshop's demand sequence D for adjusting goods and the minimum batch B input from the outer layer, first generates an initial sorting solution, then randomly searches for a new solution within the neighborhood of the solution, conducts M times of Monte Carlo simulation, compares the quality of the two solutions, and accepts the new solution according to the Metropolis criterion until the algorithm iteration reaches the maximum number of generations MaxGen, stops searching for new solutions, and outputs the current optimal sorting P B and the corresponding evaluation value C(P B ).

[0020] The initial sorting solution mentioned above refers to: on the premise of a given minimum batch size, dividing the daily demand for various vehicle models into batches, with the number of vehicles in each batch being no less than the value of the minimum batch size. First, each vehicle model is preferably divided according to the value of the minimum batch size. If there are remaining vehicles that are not enough for a batch after grouping, they are added to the last batch of the same vehicle model.

[0021] The neighborhood of the solution includes: ① Exchange: randomly exchange the positions of two items in the sequence; ② Shift: randomly intercept a segment in the sequence and insert it anywhere in the remaining sequence; ③ Inversion: randomly intercept a segment in the sequence and invert and insert it back to the original position; when the neighborhood solution does not meet the minimum batch size constraint given by the outer layer, the neighborhood solution is infeasible.

[0022] The Metropolis criterion mentioned above refers to: when the evaluation index Y of the new solution is lower than the evaluation index X of the original solution, accept the new solution; otherwise, accept the new solution with a probability where \(H = H_0\times A\) k , representing the current annealing temperature H calculated from the initial temperature \(H_0\), the annealing coefficient \(A\lt1\), and the current iteration number \(k\lt MaxGen\).

[0023] The present invention relates to a system for implementing the above method, including: a production line database module, a vehicle model retrieval order input module for final assembly, a scheduling algorithm parameter setting module, a scheduling algorithm solving module, and a solution output display module. Among them: the production line database module and the vehicle model retrieval order input module for final assembly provide problem information for the scheduling algorithm solving module. Various parameters of the optimization algorithm are set through the scheduling algorithm parameter setting module during production. The scheduling algorithm solving module is connected to the solution output display module, and converts the optimal scheduling plan sequence obtained by the solution into an intuitive and visible production sequence.

[0024] Technical effects

[0025] The present invention as a whole solves the problems in the prior art that do not consider the minimum batch size and cannot handle flexible production lines and semi-flexible batch production lines in the automotive manufacturing industry; through the combined double-layer optimization of the minimum batch size and the specific production sorting, the present invention takes the minimum batch size as a decision variable and a constraint of the inner-layer simulated annealing algorithm, rather than a random result after scheduling, and carefully considers the influence of random disturbances in production. Compared with the prior art, the present invention generally takes the minimum batch size as a dimension of the decision variable, and designs an inner-layer sorting and scheduling optimization algorithm with a minimum batch size constraint, so as to be able to perform production scheduling according to the physical characteristics of the production line. Description of the drawings

[0026] Figure 1 is the flow chart of the present invention;

[0027] Figure 2Flow chart of the Monte Carlo simulation for the embodiment. Detailed implementation manners

[0028] In this embodiment, there are 133 workstations on the production line. According to the production plan, with 1 minute as one beat, the 6 types of vehicle bodies put in are successively processed through the processes of 133 workstations and then enter the vehicle body distribution center for storage. The number of continuously produced vehicle bodies of each type is called the minimum batch quantity B (Block) of the batch. Since the existing fully flexible scheduling technology cannot perform production scheduling for this problem, B = 20 is tentatively set in the current production plan of the enterprise.

[0029] Through the production line database module and the vehicle type retrieval sequence input module for final assembly, the initial inventories of the 6 types of vehicle bodies in the vehicle body distribution center are read as 50, 10, 20, 20, 10, and 10 units respectively. The total quantity of goods transferred in one day in the final assembly workshop is N = 1200 vehicles, and the total quantities of goods transferred for the 6 types of vehicle bodies are 500, 100, 200, 200, 100, and 100 vehicles respectively. The goods transfer sequence D is: first transfer 25 vehicles of type 1, then 5 vehicles of type 2, 10 vehicles of type 3, 10 vehicles of type 4, 5 vehicles of type 5, and 5 vehicles of type 6, that is, the goods are transferred in a cycle of 60 vehicles per hour.

[0030] According to the statistical data in actual production, regarding the random event of sampling inspection, when a vehicle body enters the quality control area at each beat, the vehicle is sampled with a probability of p check = 3%, and the duration of each sampling inspection is T check = 6 beats. When inspecting, there is a probability of discovering quality problems, causing the entire workshop to stop work for repair.

[0031] According to the statistical data in actual production, regarding the random event of rework, there is a probability of p repair = 17% that rework occurs during the painting process of the vehicle body. The vehicle body is sent back to the body shop to repair the defects. The average value E(X) of the rework duration is 10 minutes (i.e., 10 beats), and the repair duration follows a lognormal distribution, that is, the parameters are μ repair = 1.8, σ repair = 1. The reworked vehicle bodies directly enter the vehicle body distribution center.

[0032] According to the statistical data in actual production, regarding the random event of workshop shutdown, the workshop has a time ratio of p malfunction = 6% in the state of production line failure. Each repair generally requires T breakdown = 60 beats. Assuming that the probability of shutdown at each beat is equal, the shutdown probability p at each beat can be deduced repair≈0.001. There are two types of faults. When the body production line stops, no sampling inspection and repair work are carried out, and no new bodies enter the body distribution center. When the final assembly line stops, the final assembly workshop does not transfer goods from the body distribution center.

[0033] As Figure 1 shown, this embodiment implements this scheduling method for the multi - model batch flexible production line in the body workshop of an automobile manufacturing enterprise, including:

[0034] Step S1, determine the evaluation coefficients of the solution: the coefficient α of the model changeover times S is 100, the coefficient β of the part transportation times T is 0.5, the coefficient δ of the part in - line storage quantity R is 0.5, and the coefficient σ of the out - of - stock times L in the body distribution center is 30. Among the evaluation indicators, about 50% is the out - of - stock times, 15% is the model changeover times, 18% is the part transportation times, and 17% is the part in - line storage quantity.

[0035] Step S2, determine the parameters of the inner - layer simulated annealing algorithm: the maximum number of iterations MaxGen = 30000, the initial temperature H0 = 1000, and the annealing coefficient A = 0.95.

[0036] Step S3, the first double - layer optimization: the possible value range of the specified minimum batch in the outer layer is (18, 60). First, within this range, it is required to perform Monte Carlo simulation M = 100 times for each solution of the inner - layer algorithm, specifically including:

[0037] Step S301, the outer layer specifies the minimum batch B.

[0038] Step S302, inner - layer initialization. According to the given minimum batch B, generate the initial sorting solution P B (0), and set the current iteration number k = 0, and the temperature H = H0 = 1000.

[0039] Step S303, evaluate the current solution, which is the weighted average C of the four indicators for P B(k)) = α·S + β·T + δ·R + σ·L, where: the number of vehicle model switches S refers to the number of times of switching vehicle models. The algorithm traverses the production plan sequence from the beginning. Each time the vehicle models produced in two consecutive batches are different, the number of switches is incremented by 1; the number of part transports T refers to the number of times of part transports. For each part, whenever a certain vehicle model is about to be produced and the corresponding part at the line side is insufficient, one or two boxes of parts will be transported according to the next usage amount of this part. Each box of parts corresponds to only one vehicle model, and at most two boxes of parts can be stored at the line side. Therefore, according to the next usage situation, the existing parts at the line side are transported back. Each time a part is transported or transported back, the number of part transports is incremented by 1; the number of parts stored at the line side R refers to the number of boxes stored at the line side (0 or 1 or 2) checked at each beat. The number of parts stored at the line side is the sum of these box numbers; the number of shortages at the vehicle body distribution center L refers to the number of shortages of transferred goods during production. When the final assembly workshop requests a transfer from the vehicle body distribution center at each beat, if the body inventory of the corresponding vehicle model is 0, one shortage is recorded.

[0040] When evaluating the solution, consider the perturbation of random events, and use Monte Carlo simulation to take the average M times. Specifically, as Figure 2 shown.

[0041] Step S304, randomly find a new solution P B (new) in the neighborhood of the three solutions of ① exchange, ② shift, and ③ inversion, and check the minimum batch constraint. If the minimum batch of the new solution does not meet the value B specified in the outer layer, find a new solution again; if it meets the minimum batch constraint, obtain the evaluation index C(P B (new)) of the new solution according to the method in Step S303.

[0042] Step S305, accept the new solution according to the Metropolis criterion: when C(P B (new)) < C(P B (k)), then accept the new solution P B (k + 1) = P B (new), otherwise accept the new solution with a probability of Accept the new solution.

[0043] Set the iteration number k = k + 1, the iteration temperature H = H * A, and repeat the process of S304 - S305 until k = MaxGen, then terminate the algorithm and output the optimal production sequence P B corresponding to the current minimum batch B and the corresponding index C(P B ).

[0044] Step S306, since the value of the solution evaluation index changes as a concave curve with the minimum batch, analyze the 10% proportion of the minimum batch values with the lowest index values, and take it as the optimal range (b3, b4) of the minimum batch.

[0045] Step S4, the second double-layer optimization. Traverse the minimum batch size within the range of (b3, b4), requiring the number of inner-layer Monte Carlo simulations M = 5000 times, and repeat the process of S3 to obtain an accurate optimal scheduling result.

[0046] Through specific actual experiments, based on the above enterprise actual production environment data, running the above method, the experimental result data that can be obtained is shown in Table 1. The solution time is 10 hours. The first double-layer optimization gives (b3, b4) = (45, 55), and the second double-layer optimization gives the optimal value of the minimum batch size B = 48. The specific vehicle model sorting is (the numbers represent the vehicle models produced in this batch): 1-1-1-1-3-3-3-3-1-4-4-4-4-1-1-1-6-1-1-2-2-5-5-6. Since there is no existing batch flexible scheduling technology, the temporarily determined plan of the enterprise is that the minimum batch size B = 20, which is roughly obtained by the table static deduction method, and the specific vehicle model sorting is also difficult to solve by the existing technology, but is obtained by using the inner-layer optimization algorithm of this embodiment when B = 20.

[0047] Table 1:

[0048]

[0049] As described above, this method first proposes a production scheduling method that can consider the minimum batch size constraint, which is a new technology never appeared in this field. Compared with the scheduling plan designed by the enterprise using the table static deduction method, the performance index improvement of this method lies in that it considers the influence of random event disturbances, and the quality of the optimal solution obtained is improved by 45.87% compared with the enterprise plan. In addition, the algorithm only takes 10 hours to solve, ensuring that the production plan for the next day can be formulated one day in advance.

[0050] The above specific implementation can be locally adjusted by those skilled in the art in different ways without departing from the principles and purposes of the present invention. The protection scope of the present invention is subject to the claims and is not limited by the above specific implementation. All implementation solutions within its scope are subject to the constraints of the present invention.

Claims

1. A scheduling method for a multi - model batch flexible production line, characterized in that, First, retrieve the configuration parameters of the production line. Considering the impacts of random events such as random sampling inspection, rework, and workshop shutdown, determine the evaluation parameters of the solution and the parameters of the inner-layer simulated annealing algorithm. Then, using the downstream vehicle model retrieval order as the input, divide the two major decision variables of the minimum batch size and vehicle model sorting into a two-layer optimization problem. Specifically: after determining the minimum batch size in the outer layer, the inner-layer simulated annealing algorithm solves the sorting and scheduling problem with the minimum batch size constraint to obtain the optimal production sorting solution. The multi-model batch flexible production line mentioned above refers to: the production line has S workstations. According to the production plan, with 1 minute as one beat, multiple vehicle model bodies are successively processed through the processes of S workstations and then enter the vehicle body distribution center for storage. The number of continuously produced bodies of each vehicle model is called the minimum batch size B of batch processing. The random sampling event mentioned above means that when a vehicle body enters the quality control area in each cycle, the vehicle is sampled with a probability of p check , and the duration of each sampling is T check cycles; when a vehicle body is sampled, the subsequent vehicle bodies are still in other processes and cannot be accelerated to replace it, which means that there will be no vehicle body output from the body shop in a certain subsequent cycle; after the sampling is completed, in order to reduce model changeovers, instead of immediately inserting into the production line, it is inserted into the nearest sequence of the same model. When inserting, the workstations before the sampling need to pause and wait; when inspecting, there is a probability of p breakdown of finding quality problems, causing the entire workshop to stop work for maintenance; The described rework random event refers to: during the painting process of the vehicle body, there is a probability of p repair that rework occurs, and the vehicle body is sent back to the body shop to repair the defects. The reworked vehicle body directly enters the vehicle body distribution center; The described random event of workshop shutdown means that the workshop has a time ratio of p malfunction in the state of production line failure, and each maintenance requires T breakdown beats; assuming the probability of shutdown for each beat is equal to p repair , then the average non-shutdown duration is The shutdown probability of each beat can be deduced There are two types of failures. When the body production line shuts down, no sampling inspection and rework are carried out, and no new bodies enter the body distribution center. When the final assembly production line shuts down, the final assembly workshop does not transfer goods from the body distribution center; The evaluation parameters of the solution mentioned above refer to: the coefficient weights of the number of vehicle model switches S, the number of part transports T, the number of parts stored at the line side R, and the number of shortages in the vehicle body distribution center L. Summing them up gives the evaluation index C of the sorting solution. The vehicle body distribution center mentioned above refers to: a vehicle body warehouse with a re-sorting function that can receive the processed vehicle bodies from the body shop, sort them according to different vehicle models, and provide the required vehicle models to the general assembly shop.

2. The scheduling method for a multi - model batch flexible production line according to claim 1, characterized in that, The downstream vehicle model retrieval order mentioned above refers to: at each beat, the general assembly shop retrieves one vehicle body of a certain vehicle model from the vehicle body distribution center for overall assembly, and this retrieval order is a pre-known sequence D with a length of N vehicle bodies.

3. The scheduling method for a multi - model batch flexible production line according to claim 1, characterized in that, The production sorting solution mentioned above refers to: a production plan sequence P with a length of N vehicle bodies, indicating the type of vehicle body put into the body shop for processing at each beat. After being processed through S workstations, the vehicle body enters the vehicle body distribution center for stockpiling.

4. The scheduling method for a multi - model batch flexible production line according to claim 1, characterized in that, The optimization objective of the two-layer optimization problem mentioned above is: minimize the evaluation index of the solution Min.C = α·S + β·T + δ·R + σ·L, where: the number of vehicle model switches S refers to: the number of times of switching vehicle models. The algorithm traverses the production plan sequence from the beginning. Whenever the vehicle models produced by two consecutive batches are different, the number of switches +1; the number of part transports T refers to: the number of part transports. For each type of part, whenever a certain vehicle model is about to be produced and the corresponding part at the line side is insufficient, it will be judged whether to transport one or two boxes of parts based on the next usage amount of this part. Each box of parts only corresponds to one vehicle model, and at most two boxes of parts can be stored at the line side. Therefore, according to the next usage situation, the existing parts at the line side are transported back. Whenever there is a part transported or transported back, the number of part transports +1; the number of parts stored at the line side R refers to: check the number of boxes stored at the line side at each beat, and the line side storage quantity is the sum of these box numbers; the number of shortages in the vehicle body distribution center L refers to: the number of shortages in stockpiling during production. When the general assembly shop requests stockpiling from the vehicle body distribution center at each beat, if the vehicle body inventory of the corresponding vehicle model is 0, record a shortage once.

5. The scheduling method for a multi - model batch flexible production line according to claim 1, characterized in that, The minimum batch size is achieved through outer-layer traversal, including roughly determining the minimum batch size range and precise search: First step, roughly determine the minimum batch size. First, determine the range of possible positive integer values of the minimum batch size from actual production: (b1, b2). Within this range, first use a smaller number of simulation runs M = M1, traverse the specified minimum batch size b1 ≤ B ≤ b2, input it into the inner simulated annealing algorithm, and obtain the corresponding optimal production sequence P B , so as to determine the optimal approximate range of the minimum batch size B Step 2: Exact search. In the range of (b3, b4), a larger number of simulation runs M = M2 > M1 is adopted, and the specified minimum batch size b3 ≤ B ≤ b4 is traversed and input into the inner simulated annealing algorithm to obtain the corresponding optimal inner production sequence P B , and the optimal evaluation value C(P B ) is selected, and the corresponding minimum batch size B and production sequence P B are output as the optimal solution.

6. The scheduling method for a multi - model batch flexible production line according to claim 1, characterized in that, The simulated annealing algorithm of the inner layer, based on the transfer demand sequence D of the general assembly workshop and the minimum batch size B input by the outer layer, first generates an initial sorting solution, then randomly searches for a new solution within the neighborhood of the solution, simulates M times through Monte Carlo simulation, compares the quality of the two solutions, accepts the new solution according to the Metropolis criterion, and stops searching for a new solution until the algorithm iteration reaches the maximum number of generations MaxGen, and outputs the current optimal sorting P B and the corresponding evaluation value C(P B ).

7. The scheduling method for a multi - model batch flexible production line according to claim 6, characterized in that, The described initial sorting solution means that, on the premise of a given minimum batch size, the demand for various vehicle models within a day is divided into batches, with the number of vehicles in each batch not less than the minimum batch size. First, each vehicle model is divided according to the minimum batch size. If there are remaining vehicles that are not enough for a full batch after grouping, they are added to the last batch of the same vehicle model.

8. The scheduling method for a multi - model batch flexible production line according to claim 6, characterized in that, The neighborhood of the described solution includes: ① Exchange: Randomly exchange the positions of two items in the sequence; ② Shift: Randomly intercept a segment of the sequence and insert it anywhere in the remaining sequence; ③ Inversion: Randomly intercept a segment of the sequence and invert it and insert it back in place. When the neighborhood solution does not meet the minimum batch size constraint given by the outer layer, this neighborhood solution is infeasible.

9. The scheduling method for a multi - model batch flexible production line according to claim 6, characterized in that, The Metropolis criterion mentioned above means that when the evaluation index Y of the new solution is lower than the evaluation index X of the original solution, the new solution is accepted; otherwise, the new solution is accepted with a probability where H = H0 × A k , and H represents the current annealing temperature calculated from the initial temperature H0, the annealing coefficient A < 1, and the current iteration number k < MaxGen.

10. A system for implementing the scheduling method for a multi - model batch flexible production line according to any one of claims 1 to 9, characterized in that, It includes: A production line database module, a vehicle model retrieval order input module for final assembly, a scheduling algorithm parameter setting module, a scheduling algorithm solving module, and a solution output display module. Among them: The production line database module and the vehicle model retrieval order input module for final assembly provide problem information for the scheduling algorithm solving module. Various parameters of the optimization algorithm are set through the scheduling algorithm parameter setting module during production. The scheduling algorithm solving module is connected to the solution output display module and converts the obtained optimal scheduling plan sequence into an intuitive and visible production sequence.

Images