Clothing production process distribution method considering hanging connectivity constraint

Abstract

The invention provides a garment production process allocation method considering hanging connectivity constraints. The method comprises the steps of obtaining station information of a production workshop, process information of a to-be-processed workpiece and connectivity information of a hanging system between stations; establishing a station distance matrix of a production workshop by taking connectivity information of the hanging system between stations as a constraint condition; generating a population of an NSGA-II algorithm by using a greedy rule according to the station distance matrix, the station information of the production workshop and the process information of the to-be-processed workpiece; constructing a multi-objective optimization model by taking the minimum total walking distance of the hanging system and the minimum average difference of the total processing working hours among all the activated stations as optimization objectives; based on the population generated in the step S3, multi-objective optimization is solved through an NSGA-II algorithm, the optimal garment production process distribution scheme is obtained, the garment production efficiency is effectively improved, and the production cost is reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of garment processing, and in particular relates to a garment production process allocation method considering hanging connectivity constraints. Background Art

[0002] In the field of modern clothing production, with the expansion of production scale and the increasing requirements of consumers for product quality and delivery time, the optimization and intelligent management of the production process have become the key to improving competitiveness. The traditional manual allocation process and scheduling method is not only inefficient, but also difficult to cope with complex and changing production needs, which easily leads to resource waste and rising production costs. Therefore, it is particularly important to develop a production process optimization allocation system based on intelligent algorithms.

[0003] In the garment production process, the workstation layout of the production workshop, the process flow of the workpieces to be processed, and the transportation efficiency of the hanging system directly affect the production efficiency and cost. The workstation layout determines the transfer path and distance of the workpieces between different processes, and the hanging system is an important transportation tool connecting the various workstations. Its connectivity and efficiency are crucial to the smoothness of the production process. However, traditional methods often ignore the interaction between these factors, resulting in unreasonable process allocation, increasing the travel distance of the hanging system and the idle waiting time of the workstations, thereby reducing the overall production efficiency and production costs. Summary of the invention

[0004] In order to solve the problems existing in the background technology, improve the efficiency of clothing production, and reduce the cost of clothing production, the present invention provides a clothing production process allocation method considering hanging connectivity constraints, including:

[0005] S1: Obtain the workstation information of the production workshop, the process information of the workpiece to be processed, and the connectivity information of the hanging system between the workstations;

[0006] S2: Use the connectivity information of the hanging system between workstations as a constraint to establish the workstation distance matrix of the production workshop;

[0007] S3: Generate the population of NSGA-II algorithm using greedy rules according to the workstation distance matrix, the workstation information of the production workshop and the process information of the workpiece to be processed;

[0008] S4: Construct a multi-objective optimization model by taking the minimum total travel distance of the hanging system and the minimum average difference in total processing time between all activated workstations as the optimization objectives;

[0009] S5: Based on the population generated in step S3, the NSGA-II algorithm is used to solve the multi-objective optimization and obtain the optimal clothing production process allocation plan.

[0010] Preferably, the establishment of the workstation distance matrix of the production workshop includes: transporting work-in-progress between different workstations through a hanging system, and constructing a workstation distance matrix according to the moving distance of the hanging system between different workstations, wherein the distance between two workstations that cannot be connected to each other is represented as -1.

[0011] Preferably, step S3 comprises:

[0012] S31: Initialize the population size to N;

[0013] S32: for each first process without a preceding process in the workpiece to be processed, randomly select a workstation from its allocatable workstation set as its allocated workstation;

[0014] S33: Taking the front-end and back-end relationship between processes as the constraint condition, traverse all processes, and select the workstation closest to the allocated workstation of the front-end process as the allocated workstation of the back-end process according to the workstation distance matrix using the greedy rule; until all processes are allocated to the allocated workstations, individuals in the NSGA-II algorithm population are generated;

[0015] S34: Repeat steps S32-S34 until the number of individuals reaches N, and obtain the population of the NSGA-II algorithm.

[0016] Preferably, the multi-objective optimization model includes:

[0017]

[0018] Among them, O1 and O2 represent two optimization goals, G represents the process set of the workpiece to be processed, l and u represent two different processes, and process u is the subsequent process of process l; k1 and k2 represent two different workstations; Y luk1k represents a binary variable, when process l is scheduled at k1 and process u is scheduled at k2, Y luk1k2 is equal to 1, otherwise Y luk1k =0; D k1k2 K represents the distance between station k1 and station k2 in the station distance matrix; ava Indicates the total number of all activated workstations; CA k represents the total processing time of station k, Indicates the average processing time of all activated workstations.

[0019] Preferably, step S5 comprises:

[0020] S51: Perform non-dominated sorting on the population of the NSGA-II algorithm to obtain the rank of each individual in the population;

[0021] S52: traverse each optimization target in the multi-objective optimization model, calculate the crowding degree of each individual in the population under the optimization target, and sum up the crowding degree of each individual under all optimization targets to obtain the final crowding degree of each individual in the population;

[0022] S53: Select X individuals from the population as elite individuals for retention using the elite retention strategy according to the individual's level and the individual's final crowding degree, and select parent individuals from the population using the tournament selection strategy to generate a parent population;

[0023] S54: Generate a child population using a simulated binary crossover and mutation algorithm for the parent population, and merge the parent population and the child population to form a temporary population;

[0024] S55: select Y individuals from the temporary population according to the preset probability model, X+Y=N, N represents the size of the population;

[0025] S56: The individuals screened out in step S55 and the retained elite individuals are combined to form the population for the next iteration, and steps S51-S56 are repeated until the preset number of iterations is reached, and the optimal solution is output.

[0026] Preferably, the non-dominated sorting of the population of the NSGA-II algorithm comprises:

[0027] S511: For each individual P in the population, initialize the set S p and counter n p ;

[0028] S512: For each individual p in the population, traverse the other individuals q in the population. If individual p dominates individual q, add individual q to the set S. p If individual p is dominated by individual q, then n p Add 1;

[0029] S513: Traverse the individuals in the population, if there is a counter n of individual p p When it is 0, individual p is placed in the non-dominated frontier set F1, and the rank of individual p is set to 0;

[0030] S514: For each non-dominated front set F i , traverse the non-dominated frontier set F i For each individual p in the set S p For each individual q in the q Minus 1, if n q If it decreases to 0, the rank of individual q is set to i+1, and individual q is added to the next non-dominated frontier set F i+1 ;

[0031] S515: When the non-dominated frontier set is empty, the sorting process ends and the rank of each individual in the population is obtained.

[0032] Preferably, the step of selecting X individuals from a population as elite individuals using an elite retention strategy comprises:

[0033] S5211: Calculate the population size as N, the number of elite individuals retained X = [α·N] 取整 , α is a real number in the interval (0, 1];

[0034] S5212: Starting from the first non-dominated frontier set F1, individuals are selected from the non-dominated frontier set as elite individuals in sequence until a predetermined number of elite individuals X is reached; the selection strategy is as follows:

[0035] First, select all individuals in the non-dominated frontier set F1. If the number of individuals in set F1 is less than X, continue to select from the non-dominated frontier set F2, and so on. If the number of elite individuals selected reaches X or all non-dominated frontier sets have been traversed, stop selecting. If in a non-dominated frontier set F i Some individuals need to be selected as elite individuals, then according to the set F i The final crowding degree of individuals in the pool is selected from high to low.

[0036] Preferably, the crowding degree of each individual under the optimization objective includes:

[0037]

[0038] Where m∈{O1,O2} represents the optimization target; Im(i) represents the crowding degree of individual i under the optimization target m; f m (i+1) represents the value of individual i+1 under the optimization target m; f m (i-1) represents the value of individual i-1 under the optimization target m; Represents the maximum value of all individuals under the optimization target m; Represents the minimum value of all individuals under the optimization target m.

[0039] Preferably, the step of selecting parent individuals from a population using a tournament selection strategy to generate a parent population comprises:

[0040] S5221: Randomly select k individuals from the current population to form a tournament group;

[0041] S5222: In the tournament population, by comparing the individual levels and final crowding, select the best individual as the parent individual. The selection strategy is as follows:

[0042] For individuals p and q, if the rank(p) of individual p is less than the rank(q) of individual q, then select individual p; if the rank(p) of individual p is greater than the rank(q) of individual q, then select individual q; if the rank(p) of individual p is equal to the rank(q) of individual q, then select the individual with the larger final crowding degree between individuals p and q; if the final crowding degrees of individuals p and q are also the same, then randomly select an individual from individuals p and q as the parent individual;

[0043] S5223: Repeat steps S5221 to S5223 until the number of parent individuals reaches a set threshold to generate a parent population.

[0044] Preferably, the generating the offspring population by using a simulated binary crossover and mutation algorithm on the parent population comprises:

[0045] Simulating binary crossover selection includes: randomly selecting two parent individuals from the parent population, randomly selecting some genes from the two parent individuals for exchange, and generating two offspring individuals;

[0046] The mutation algorithm selection includes: randomly selecting a parent individual from the parent population, and changing the workstation assigned to a randomly selected process from the parent to generate a child individual;

[0047] Based on the connectivity and workstation information of the hanging system, determine whether the generated offspring individuals are reasonable. If not, discard the unreasonable offspring individuals.

[0048] A preset number of offspring populations are generated based on the above-mentioned simulated binary crossover and / or mutation algorithm.

[0049] The present invention has at least the following beneficial effects

[0050] The present invention reduces the total travel distance of the hanging system between workstations by optimizing the process allocation, directly reduces the time cost of material transportation, and speeds up the production process. At the same time, the average difference in processing time between each workstation is balanced, avoiding the situation where some workstations are idle for a long time or are overly busy, thereby improving the overall production efficiency. Reducing the travel distance of the hanging system means reducing energy consumption and equipment wear, which helps to reduce maintenance costs and energy consumption in the long run. In addition, by optimizing the process allocation, production stagnation caused by waiting or congestion is reduced, further reducing production costs. The scheme takes into account the actual environment of the production workshop and the connectivity of the hanging system, making the process allocation more flexible and adaptable. When the production demand or workshop layout changes, the optimal process allocation scheme can be regenerated by adjusting the parameters in the optimization model to ensure the continuous and efficient production process. The scheme introduces an intelligent algorithm to optimize the process allocation, which provides strong support for the intelligent management of clothing production enterprises. Through data-driven and algorithm optimization, enterprises can grasp the production status more accurately and realize real-time monitoring and dynamic adjustment of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0052] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0053] See also Figure 1 The present invention provides a method for allocating garment production processes taking into account the constraints of hanging connectivity, comprising:

[0054] S1: Obtain the workstation information of the production workshop, the process information of the workpiece to be processed, and the connectivity information of the hanging system between the workstations;

[0055] S2: Use the connectivity information of the hanging system between workstations as a constraint to establish the workstation distance matrix of the production workshop;

[0056] S3: Generate the population of NSGA-II algorithm using greedy rules according to the workstation distance matrix, the workstation information of the production workshop and the process information of the workpiece to be processed;

[0057] S4: Construct a multi-objective optimization model by taking the minimum total travel distance of the hanging system and the minimum average difference in total processing time between all activated workstations as the optimization objectives;

[0058] S5: Based on the population generated in step S3, the NSGA-II algorithm is used to solve the multi-objective optimization and obtain the optimal clothing production process allocation plan.

[0059] In this embodiment, the workstation information includes the workstation ID, the workstation location and the process set that can be processed by the workstation, etc.; the process information of the workpiece to be processed includes: all processes of the workpiece to be processed, the standard working time of the process, and the process set of the subsequent processes, etc. The relationship between the processes is expressed as one process must be started after the completion of another process, which is a common expression method in this field. The workpieces are transported through the connectivity of the hanging system between different workstations;

[0060] By taking the minimum total travel distance of the hanging system and the minimum average difference in total processing time between all activated workstations as the optimization goals, the total travel distance of the hanging system between workstations is reduced, which directly reduces the time cost of material transportation, speeds up the production process, balances the average difference in processing time between workstations, avoids the situation where some workstations are idle for a long time or are overly busy, thereby improving overall production efficiency.

[0061] Preferably, the establishment of the workstation distance matrix of the production workshop includes: transporting work-in-progress between different workstations through a hanging system, and constructing a workstation distance matrix according to the moving distance of the hanging system between different workstations, wherein the distance between two workstations that cannot be connected to each other is represented as -1.

[0062] In this embodiment, by accurately measuring and recording the actual moving distance of the hanging system between different workstations and constructing a workstation distance matrix, a more accurate data basis is provided for the subsequent process allocation optimization. This helps the algorithm to more accurately evaluate the transportation costs under different process allocation schemes, thereby generating optimization results that are closer to actual needs. The distance between workstations that cannot be connected to each other is set to -1. This processing method effectively reflects the actual layout of the production workshop and the connectivity limitations of the hanging system. This makes the optimization model closer to production reality, avoids the generation of unfeasible process allocation schemes, and enhances the applicability and practicality of the model.

[0063] Preferably, step S3 comprises:

[0064] S31: Initialize the population size to N;

[0065] S32: for each first process without a preceding process in the workpiece to be processed, randomly select a workstation from its allocatable workstation set as its allocated workstation;

[0066] S33: Taking the front-end and back-end relationship between processes as the constraint condition, traverse all processes, and select the workstation closest to the allocated workstation of the front-end process as the allocated workstation of the back-end process according to the workstation distance matrix using the greedy rule; until all processes are allocated to the allocated workstations, individuals in the NSGA-II algorithm population are generated;

[0067] S34: Repeat steps S32-S34 until the number of individuals reaches N, and obtain the population of the NSGA-II algorithm.

[0068] In this embodiment, the number of processes m = 6 and the number of workstations s = 4 are taken as an example. The processes o1 to o6 represent the order of processing according to the sequence numbers from small to large. Since the machines of the workstations have been determined, the workstations that can be assigned to each process have also been determined. Table 1 shows the workstations that can be arranged for each process. This example also assumes that the four workstations are all connected, and the distance between them increases with the increase of the sequence number.

[0069] Table 1 The workstations that can be arranged for each process

[0070]

[0071] Search for the first process; that is, search for the process without a preceding process among all process pairs that have a sequential relationship. In this example, o1 is the first process;

[0072] Starting from the first process o1, randomly select s1 from its allocatable workstation set [s1, s2] as its final assigned workstation. Taking the relationship between the previous and next processes of the workpiece as the constraint condition, its subsequent process is o2. Based on the allocatable workstation set of its subsequent process o2, the workstation selection is performed according to the greedy rule.

[0073] For the set of workstations [s3, s4] that can be assigned to the post-process of the first process o1, assuming that the distance matrix of each workstation is dis, the workstation closest to o2 is greedily selected based on the distance between the selectable workstation o2 and the workstation s1. The selection formula is as follows:

[0074] sel_station=min(dis[1,2],dis[1,4])

[0075] After determining the workstation for process o2, repeat the above steps until all processes are assigned to workstations. For this example, the final plan is (s1, s2, s3, s4, s4, s4).

[0076] In this embodiment, before executing step S32, it can also include: searching the historical database, going through each allocation plan stored in the historical database, and determining whether the process information of each plan is consistent with the process information of the workpiece to be processed. If so, the allocation plan in the historical database is treated as an individual, and step S32 is executed. After executing step S5, the obtained allocation plan is stored in the historical database.

[0077] In this embodiment, the first process is randomly selected to allocate workstations (S32), and based on this, the greedy rule (S33) is used to gradually allocate workstations for the subsequent processes until all processes are allocated. This method can quickly generate the initial population. The convergence speed is improved, the algorithm calculation amount is reduced, and the sensitivity and response speed of the algorithm are enhanced. This scheme has shown significant beneficial effects in terms of accuracy, efficiency, resource utilization, solution diversity, subsequent optimization support, and algorithm convergence speed.

[0078] Preferably, the multi-objective optimization model includes:

[0079]

[0080] Among them, O1 and O2 represent two optimization goals, G represents the process set of the workpiece to be processed, l and u represent two different processes, and process u is the subsequent process of process l; k1 and k2 represent two different workstations; Y luk1k represents a binary variable, when process l is scheduled at k1 and process u is scheduled at k2, Y luk1k2 is equal to 1, otherwise Y luk1k =0; D k1k2 K represents the distance between station k1 and station k2 in the station distance matrix; ava Indicates the total number of all activated workstations; CA k represents the total processing time of station k, Indicates the average processing time of all activated workstations.

[0081] In this embodiment, the total travel distance of the suspension system between workstations is reduced by taking the minimum total travel distance of the suspension system and the minimum average difference in total processing time between all activated workstations as optimization goals, thereby directly reducing the time cost of material transportation, speeding up the production process, balancing the average difference in processing time between each workstation, and avoiding the situation where some workstations are idle for a long time or are overly busy, thereby improving overall production efficiency.

[0082] Preferably, step S5 comprises:

[0083] S51: Perform non-dominated sorting on the population of the NSGA-II algorithm to obtain the rank of each individual in the population;

[0084] S52: traverse each optimization target in the multi-objective optimization model, calculate the crowding degree of each individual in the population under the optimization target, and sum up the crowding degree of each individual under all optimization targets to obtain the final crowding degree of each individual in the population;

[0085] S53: Select X individuals from the population as elite individuals for retention using the elite retention strategy according to the individual's level and the individual's final crowding degree, and select parent individuals from the population using the tournament selection strategy to generate a parent population;

[0086] S54: Generate a child population using a simulated binary crossover and mutation algorithm for the parent population, and merge the parent population and the child population to form a temporary population;

[0087] S55: select Y individuals from the temporary population according to the preset probability model, X+Y=N, N represents the size of the population;

[0088] S56: The individuals screened out in step S55 and the retained elite individuals are combined to form the population for the next iteration, and steps S51-S56 are repeated until the preset number of iterations is reached, and the optimal solution is output.

[0089] In this embodiment, step S51 is sorted by non-dominated sorting, and the algorithm can distinguish the superiority and inferiority relationship between different individuals in the population and divide the individuals into different levels (frontiers). This helps to give priority to better individuals in the subsequent selection process, thereby guiding the algorithm to approach the global optimal solution. Crowding is an indicator to measure the density of the surrounding solutions of an individual in the population. By calculating the crowding degree of each individual under all optimization objectives and summing up to obtain the final crowding degree, the algorithm can further distinguish the superiority and inferiority among individuals of the same level and maintain the diversity of the population. This helps the algorithm avoid falling into the local optimal solution during the search process. The elite retention strategy directly retains the excellent individuals in the current population to the next generation, avoiding the loss of excellent solutions. This helps the algorithm to continuously accumulate excellent solutions during the iteration process, improve the convergence speed of the algorithm and the quality of the solution. Tournament selection is a selection strategy based on random competition, which selects parent individuals by comparing the fitness of a few individuals. This method can maintain the diversity of the population to a certain extent and help the algorithm jump out of the local optimal solution. Merging the parent population and the child population to form a temporary population helps the algorithm to search for the optimal solution in a larger solution space. According to the preset probability model, a certain number of individuals are selected from the temporary population to keep the size of the population unchanged. This process can further filter out poor solutions and retain better solutions.

[0090] In this embodiment, the preset probability model includes: a random selection mechanism (roulette selection) or a deterministic selection mechanism (tournament selection strategy).

[0091] Preferably, the non-dominated sorting of the population of the NSGA-II algorithm comprises:

[0092] S511: For each individual P in the population, initialize the set S p and counter n p ;

[0093] S512: For each individual p in the population, traverse the other individuals q in the population. If individual p dominates individual q, add individual q to the set S. p If individual p is dominated by individual q, then n p Add 1;

[0094] S513: Traverse the individuals in the population, if there is a counter n of individual p p When it is 0, individual p is placed in the non-dominated frontier set F1, and the rank of individual p is set to 0;

[0095] S514: For each non-dominated front set F i , traverse the non-dominated frontier set F i For each individual p in the set S p For each individual q in the q Minus 1, if n q If it decreases to 0, the rank of individual q is set to i+1, and individual q is added to the next non-dominated frontier set F i+1 ;

[0096] S515: When the non-dominated frontier set is empty, the sorting process ends and the rank of each individual in the population is obtained.

[0097] Preferably, the step of selecting X individuals from a population as elite individuals using an elite retention strategy comprises:

[0098] S5211: Calculate the population size as N, the number of elite individuals retained X = [α·N] 取整 , α is a real number in the interval (0, 1];

[0099] S5212: Starting from the first non-dominated frontier set F1, individuals are selected from the non-dominated frontier set as elite individuals in sequence until a predetermined number of elite individuals X is reached; the selection strategy is as follows:

[0100] First, select all individuals in the non-dominated frontier set F1. If the number of individuals in set F1 is less than X, continue to select from the non-dominated frontier set F2, and so on. If the number of elite individuals selected reaches X or all non-dominated frontier sets have been traversed, stop selecting. If in a non-dominated frontier set Fi Some individuals need to be selected as elite individuals, then according to the set F i The final crowding degree of individuals in the pool is selected from high to low.

[0101] Preferably, the crowding degree of each individual under the optimization objective includes:

[0102]

[0103] Where m∈{O1,O2} represents the optimization objective; I m (i) represents the crowding degree of individual i under the optimization target m; f m (i+1) represents the value of individual i+1 under the optimization target m; f m (i-1) represents the value of individual i-1 under the optimization target m; Represents the maximum value of all individuals under the optimization target m; Represents the minimum value of all individuals under the optimization target m.

[0104] Preferably, the step of selecting parent individuals from a population using a tournament selection strategy to generate a parent population comprises:

[0105] S5221: Randomly select k individuals from the current population to form a tournament group;

[0106] S5222: In the tournament population, by comparing the individual levels and final crowding, select the best individual as the parent individual. The selection strategy is as follows:

[0107] For individuals p and q, if the rank(p) of individual p is less than the rank(q) of individual q, then select individual p; if the rank(p) of individual p is greater than the rank(q) of individual q, then select individual q; if the rank(p) of individual p is equal to the rank(q) of individual q, then select the individual with the larger final crowding degree between individuals p and q; if the final crowding degrees of individuals p and q are also the same, then randomly select an individual from individuals p and q as the parent individual;

[0108] S5223: Repeat steps S5221 to S5223 until the number of parent individuals reaches a set threshold to generate a parent population.

[0109] Preferably, the generating the offspring population by using a simulated binary crossover and mutation algorithm on the parent population comprises:

[0110] Simulating binary crossover selection includes: randomly selecting two parent individuals from the parent population, randomly selecting some genes from the two parent individuals for exchange, and generating two offspring individuals;

[0111] The mutation algorithm selection includes: randomly selecting a parent individual from the parent population, changing a randomly selected process from the parent to generate a child individual by changing its assigned workstation; judging whether the generated child individual is reasonable based on the connectivity of the hanging system and the workstation information, and discarding the unreasonable child individual if it is unreasonable;

[0112] A preset number of offspring populations are generated based on the above-mentioned simulated binary crossover and / or mutation algorithm.

[0113] In this embodiment, the probability ratio of binary crossover and mutation algorithm is selected to be 0.9:0.1;

[0114] Please refer to the assumption in this embodiment that the two parent chromosomes are p1 = [2, 2, 2, 3, 3, 5, 6, 7, 8] and p2 = [1, 1, 2, 3, 4, 5, 6, 6, 8]. When performing the crossover operation, the starting and ending positions are first randomly selected, such as 2 and 4. Then, the 2nd to 4th segments of the two are exchanged, and the new daughter chromosomes obtained are c1 = [2, 2, 2, 3, 4, 5, 6, 7, 8] and c2 = [1, 1, 2, 3, 3, 5, 6, 6, 8].

[0115] Chromosome mutation: Assume that the parent chromosome is d1[2, 2, 2, 3, 3, 5, 6, 7, 8], randomly select a position, and then change the arrangement of the process. For example, if the selected number is 4, it means that process 4 is arranged at station 3. Assuming that process 4 can also be arranged at station 2, the mutation method at this time is to change 3 to 2

[0116] Assuming that the parent chromosome is p1 = [2,2,2,3,3,5,6,7,8], when performing the mutation operation, a position is randomly selected and the arrangement of the workstation of the process is changed. For example, the selected sequence number is 3, which means that process 4 is arranged at workstation 3. Assuming that process 4 can also be arranged at workstation 2, the mutation method is to adjust the workstation arrangement of process 4 from 3 to 2. Through the above mutation operation, the new mutation chromosome obtained is m1 = [2,2,2,2,3,5,6,7,8].

[0117] In this embodiment, the encoding of an individual is represented by {a1, a2, ..., a M}, M represents the serial number of the workstation, a M Represents the set of processes assigned to a workstation, represented by a tensor.

[0118] In this embodiment, judging whether the crossover is reasonable based on the connectivity of the hanging system and the workstation information includes:

[0119] Determine whether the workstations where the previous and subsequent processes of each process assigned to the workstation are connected to the workstation, and whether each workstation can process the assigned process. If so, the offspring individuals are reasonable, otherwise they are unreasonable.

[0120] In this embodiment, the optimal solution is represented by the individual with the smallest rank. If the ranks of two individuals are the same, the individual with a larger congestion degree is the optimal solution.

[0121] In summary, the present invention reduces the total distance traveled by the hanging system between workstations by optimizing the process allocation, directly reduces the time cost of material transportation, and speeds up the production process. At the same time, the average difference in processing time between each workstation is balanced, avoiding the situation where some workstations are idle for a long time or are overly busy, thereby improving the overall production efficiency. Reducing the travel distance of the hanging system means reducing energy consumption and equipment wear, which helps to reduce maintenance costs and energy consumption in the long run. In addition, by optimizing the process allocation, production stagnation caused by waiting or congestion is reduced, further reducing production costs. The scheme takes into account the actual environment of the production workshop and the connectivity of the hanging system, making the process allocation more flexible and adaptable. When the production demand or workshop layout changes, the optimal process allocation scheme can be regenerated by adjusting the parameters in the optimization model to ensure the continuous and efficient production process. The scheme introduces an intelligent algorithm to optimize the process allocation, which provides strong support for the intelligent management of clothing production enterprises. Through data-driven and algorithm optimization, enterprises can grasp the production status more accurately and realize real-time monitoring and dynamic adjustment of the production process.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A method for allocating garment production processes considering hanging connectivity constraints, characterized in that: include: S1: Obtain the workstation information of the production workshop, the process information of the workpiece to be processed, and the connectivity information of the hanging system between the workstations; S2: Use the connectivity information of the hanging system between workstations as a constraint to establish the workstation distance matrix of the production workshop; S3: Generate the population of NSGA-II algorithm using greedy rules according to the workstation distance matrix, the workstation information of the production workshop and the process information of the workpiece to be processed; S4: The multi-objective optimization model is constructed by taking the minimum total travel distance of the hanging system and the minimum average difference in total processing time between all activated workstations as the optimization objectives; S5: Based on the population generated in step S3, the NSGA-II algorithm is used to solve the multi-objective optimization and obtain the optimal clothing production process allocation plan.

2. A method for allocating garment production processes considering hanging connectivity constraints according to claim 1, characterized in that: The establishment of the workstation distance matrix of the production workshop includes: transporting work-in-progress between different workstations through a hanging system, and constructing a workstation distance matrix according to the moving distance of the hanging system between different workstations, wherein the distance between two workstations that cannot be connected to each other is represented as -1.

3. The method for allocating garment production processes considering hanging connectivity constraints according to claim 1, characterized in that: The step S3 comprises: S31: Initialize the population size to N; S32: for each first process without a preceding process in the workpiece to be processed, randomly select a workstation from its allocatable workstation set as its allocated workstation; S33: Taking the front-end and back-end relationship between processes as the constraint condition, traverse all processes, and select the workstation closest to the allocated workstation of the front-end process as the allocated workstation of the back-end process according to the workstation distance matrix using the greedy rule; until all processes are allocated to the allocated workstations, individuals in the NSGA-II algorithm population are generated; S34: Repeat steps S32-S34 until the number of individuals reaches N, and obtain the population of the NSGA-II algorithm.

4. The method for allocating garment production processes considering hanging connectivity constraints according to claim 1, characterized in that: The multi-objective optimization model includes: Among them, O1 and O2 represent two optimization goals, G represents the process set of the workpiece to be processed, l and u represent two different processes, and process u is the subsequent process of process l; k1 and k2 represent two different workstations; Y luk1k represents a binary variable, when process l is scheduled at k1 and process u is scheduled at k2 luk1k is equal to 1, otherwise Y luk1 =0; D k1k2 K represents the distance between station k1 and station k2 in the station distance matrix; ava Indicates the total number of all activated workstations; CA k represents the total processing time of station k, Indicates the average processing time of all activated workstations.

5. The method for allocating garment production processes considering hanging connectivity constraints according to claim 1, characterized in that: The step S5 comprises: S51: Perform non-dominated sorting on the population of the NSGA-II algorithm to obtain the rank of each individual in the population; S52: traverse each optimization target in the multi-objective optimization model, calculate the crowding degree of each individual in the population under the optimization target, and sum up the crowding degree of each individual under all optimization targets to obtain the final crowding degree of each individual in the population; S53: Select X individuals from the population as elite individuals for retention using the elite retention strategy according to the individual's level and the individual's final crowding degree, and select parent individuals from the population using the tournament selection strategy to generate a parent population; S54: using a simulated binary crossover and / or mutation algorithm on the parent population to generate a child population, and merging the parent population and the child population to form a temporary population; S55: select Y individuals from the temporary population according to the preset probability model, X+Y=N, N represents the size of the population; S56: The individuals screened out in step S55 and the retained elite individuals are combined to form the population for the next iteration, and steps S51-S56 are repeated until the preset number of iterations is reached, and the optimal solution is output.

6. A method for allocating garment production processes considering hanging connectivity constraints according to claim 5, characterized in that: The non-dominated sorting of the population of the NSGA-II algorithm includes: S511: For each individual P in the population, initialize the set S p and counter n p ; S512: For each individual p in the population, traverse the other individuals q in the population. If individual p dominates individual q, add individual q to the set S. p If individual p is dominated by individual q, then n p Add 1; S513: Traverse the individuals in the population, if there is a counter n of individual p p When it is 0, individual p is placed in the non-dominated frontier set F1, and the rank of individual p is set to 0; S514: For each non-dominated front set F i , traverse the non-dominated frontier set F i For each individual p in the set S p For each individual q in the q Minus 1, if n q If it decreases to 0, the rank of individual q is set to i+1, and individual q is added to the next non-dominated frontier set F i+1 ; S515: When the non-dominated frontier set is empty, the sorting process ends and the rank of each individual in the population is obtained.

7. A method for allocating garment production processes considering hanging connectivity constraints according to claim 6, characterized in that: The method of selecting X individuals from a population as elite individuals by using an elite retention strategy includes: S5211: Calculate the population size as N, the number of elite individuals retained X = [α·N] 取整 , α is a real number in the interval (0,1]; S5212: Starting from the first non-dominated frontier set F1, individuals are selected from the non-dominated frontier set as elite individuals in sequence until a predetermined number of elite individuals X is reached; the selection strategy is as follows: First, select all individuals in the non-dominated frontier set F1. If the number of individuals in set F1 is less than X, continue to select from the non-dominated frontier set F2, and so on. If the number of elite individuals selected reaches X or all non-dominated frontier sets have been traversed, stop selecting. If in a non-dominated frontier set F i Some individuals need to be selected as elite individuals, then according to the set F i The final crowding degree of individuals in the pool is selected from high to low.

8. The method for allocating garment production processes considering hanging connectivity constraints according to claim 5, characterized in that: The crowding degree of each individual under the optimization objective includes: Where m∈{O1,O2} represents the optimization objective; I m (i) represents the crowding degree of individual i under the optimization target m; f m (i+1) represents the value of individual i+1 under the optimization target m; f m (i-1) represents the value of individual i-1 under the optimization target m; Represents the maximum value of all individuals under the optimization target m; Represents the minimum value of all individuals under the optimization target m.

9. The method for allocating garment production processes considering hanging connectivity constraints according to claim 5, characterized in that: The method of selecting parent individuals from a population using a tournament selection strategy to generate a parent population includes: S5221: Randomly select k individuals from the current population to form a tournament group; S5222: In the tournament population, by comparing the individual levels and final crowding, select the best individual as the parent individual. The selection strategy is as follows: For individuals p and q, if the rank(p) of individual p is less than the rank(q) of individual q, then select individual p; if the rank(p) of individual p is greater than the rank(q) of individual q, then select individual q; if the rank(p) of individual p is equal to the rank(q) of individual q, then select the individual with the larger final crowding degree between individuals p and q; if the final crowding degrees of individuals p and q are also the same, then randomly select an individual from individuals p and q as the parent individual; S5223: Repeat steps S5221 to S5223 until the number of parent individuals reaches a set threshold to generate a parent population.

10. The method for allocating garment production processes considering hanging connectivity constraints according to claim 5, characterized in that: The method of generating a child population by using a simulated binary crossover and / or mutation algorithm on a parent population comprises: Simulating binary crossover selection includes: randomly selecting two parent individuals from the parent population, randomly selecting some genes from the two parent individuals for exchange, and generating two offspring individuals; The mutation algorithm selection includes: randomly selecting a parent individual from the parent population, and changing the workstation assigned to a randomly selected process from the parent to generate a child individual; Based on the connectivity and workstation information of the hanging system, determine whether the generated offspring individuals are reasonable. If not, discard the unreasonable offspring individuals. A preset number of offspring populations are generated based on the above-mentioned simulated binary crossover and / or mutation algorithm.

Images