Intelligent scheduling method for integration of production and transportation in integrated manufacturing system

Abstract

The application relates to the technical field of reconfigurable distributed flow shop group production and transportation integrated scheduling, in particular to an intelligent scheduling method for integrated manufacturing system production and transportation integration. The method comprises the following steps: initializing parameters, initializing a population and a current optimal solution, generating a solution of the population and a current optimal solution according to a descending order arrangement rule of processing time; updating the current optimal solution through outer loop variable neighborhood descent search, wherein the outer loop variable neighborhood descent search is embedded with inner loop variable neighborhood descent search and collaborative search based on sequential crossover and memory mechanism; population restart; outputting the current optimal solution when a termination condition is reached, or continuing to search otherwise. The application solves the technical problem that the existing method is difficult to balance equipment load and transportation efficiency in a complex production and transportation closely interwoven scene, and has the positive effects of improving overall production efficiency, optimizing resource utilization and reducing production cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of reconfigurable distributed flow shop group production and transportation integrated scheduling, and particularly belongs to an intelligent scheduling method for production and transportation integration in an integrated manufacturing system. BACKGROUND

[0002] In the current production context, with the increasing demand for efficient, flexible and intelligent manufacturing, traditional production modes face many challenges. Especially in complex multi-process, multi-factory processing environment, how to effectively coordinate resources, improve production efficiency and reduce costs has become a problem for manufacturing enterprises to solve. To address this challenge, more and more enterprises are introducing flexible scheduling and resource allocation methods to continuously optimize production processes, thereby improving overall production efficiency and promoting the development of manufacturing towards more intelligent and efficient direction.

[0003] With the rapid development of manufacturing, the problem of reconfigurable distributed flow shop group production and transportation integrated scheduling has gradually become a research hotspot. This problem involves coordination between factories, dynamic scheduling of groups and optimization of transportation links. With the high integration of production and logistics, traditional optimization methods often cannot effectively combine equipment load balancing and transportation efficiency, resulting in insufficient resource utilization. Therefore, it is necessary to break through the existing limitations and promote the deep collaboration between manufacturing processes and logistics scheduling, thereby improving production capacity and flexibility. SUMMARY

[0004] The present application provides an intelligent scheduling method for production and transportation integration in an integrated manufacturing system, which solves the technical problem that existing methods cannot balance equipment load and transportation efficiency in complex production and transportation scenarios, so as to improve overall production efficiency, optimize resource utilization and reduce production cost.

[0005] The intelligent scheduling method for production and transportation integration in an integrated manufacturing system provided by the present application is characterized by comprising the following steps,

[0006] S1, analyze the characteristics of the reconfigurable distributed flow shop group production and transportation integrated scheduling problem in the double-sided mounting process of printed circuit boards, determine the minimum maximum completion time as the solving objective, and initialize parameters including population size P , maximum search number of outer loop variable neighborhood descent O , maximum search number of inner loop variable neighborhood descent I , maximum number of coordination C , maximum number of solution updates of the population L ;

[0007] S2, initialize the population and the current optimal solution, and generate the population according to the descending order rule of processing timeP one of the solutions and one of the current optimal solutions;

[0008] S3, updating the current optimal solution by outer loop variable neighborhood descent search, wherein the outer loop variable neighborhood descent search is embedded with inner loop variable neighborhood descent search and cooperative search based on sequential crossover and memory mechanism;

[0009] S4, if the number of consecutive failures of updating the solutions of the population by the outer loop variable neighborhood descent search and the inner loop variable neighborhood descent search reaches the maximum number of updating failures of the solutions of the population L , performing population restart;

[0010] S5, judging whether the termination condition has been reached, if yes, the algorithm ends and outputs the current optimal solution, otherwise, returning to S3 to continue searching.

[0011] Further, the population is represented by , each solution is composed of two parts, represents the th solution of the population, and each solution contains the workpiece family sequence of the B surface process and the A surface process of the printed circuit board, is the sequence part, and are one-dimensional vectors, is a two-dimensional vector, n represents the number of workpiece families, represents the workpiece family sequence of the B surface process, is the workpiece family sequence of the A surface process, is the workpiece sequence in the workpiece family, is a one-dimensional vector, representing the workpiece sequence in the workpiece family ; is the allocation part, and are two-dimensional vectors, represents the selected processing cell allocation scheme of the workpiece family for the B surface process and the A surface process, represents the automated guided vehicle allocation scheme of the transportation stage of the B surface process and the A surface process of the workpiece family.

[0012] Further, the sequence part of the solutions of the population and the current optimal solution is generated according to the descending order of processing time, and the descending order of processing time is as follows,

[0013] The processing time of each workpiece in the workpiece family is compared, and the workpieces are arranged in descending order according to the processing time to generate the workpiece sequence in the workpiece family. The processing time of the workpieces in the workpiece family is summed up as the processing time of the workpiece family, and the workpiece families are arranged in descending order according to the processing time of the workpiece families to generate the workpiece family sequence.

[0014] Further, the outer loop variable neighborhood descent search adopts an outer loop neighborhood structure to traverse and optimize the population, and the operation process of the outer loop neighborhood structure includes,

[0015] job family exchange, randomly selecting a job family sequence of one process from the B-side process and the A-side process, randomly selecting two different job families from the selected job family sequence, and exchanging the positions of the selected two job families in the job family sequence;

[0016] job family insertion, randomly selecting a job family sequence of one process from the B-side process and the A-side process, randomly selecting a job family and a position in the job family sequence, and inserting the selected job family into the selected position in the job family sequence;

[0017] job exchange, randomly selecting a job family, randomly selecting two different jobs from the job sequence in the job family, and exchanging the positions of the selected two jobs in the job sequence;

[0018] job insertion, randomly selecting a job family, randomly selecting a job and a position in the job sequence in the job family, and inserting the selected job into the selected position in the job sequence;

[0019] The execution order of the operation of the outer loop neighborhood structure is job family insertion, job family exchange, job insertion, and job exchange.

[0020] Further, the inner loop variable neighborhood descent search adopts a key path-based inner loop neighborhood structure to traverse and optimize the population, and the operation process of the inner loop neighborhood structure includes,

[0021] key job family exchange, randomly selecting a job family sequence of one process from the B-side process and the A-side process, randomly selecting two different key job families on the key path from the selected job family sequence, and exchanging the positions of the two key job families in the job family sequence;

[0022] key and non-key job family exchange, randomly selecting a job family sequence of one process from the B-side process and the A-side process, randomly selecting a key job family on the key path and a normal job family on the non-key path from the selected job family sequence, and exchanging the positions of the selected key job family and the normal job family in the job family sequence;

[0023] key job family, randomly select two different jobs from the selected key job family, and exchange the positions of the selected two jobs in the job sequence of the key job family;

[0024] key job family, randomly select a job and a position in the job sequence from the selected key job family, and insert the selected job into the selected position in the job sequence of the key job family;

[0025] processing unit mutation, randomly select a key job family on the critical path, randomly select the B-side process or the A-side process of the key job family, if the B-side process of the key job family is selected, replace the processing unit allocated by the B-side process of the key job family with a random processing unit, if the A-side process of the key job family is selected, replace the processing unit allocated by the A-side process of the key job family with a random processing unit;

[0026] processing unit exchange, randomly divide the processing units allocated by the B-side process and the A-side process of the key job families and the key job families into two sets, respectively, processing unit set and processing unit set , processing unit set is used as the processing unit set allocated by the B-side process and the A-side process of the key job families , and processing unit set is used as the processing unit set allocated by the B-side process and the A-side process of the key job families ;

[0027] automated guided vehicle mutation, randomly select a key job family on the critical path, randomly select the transportation stage of the B-side process or the A-side process of the key job family, if the transportation stage of the B-side process of the key job family is selected, replace the automated guided vehicle allocated by the transportation stage of the B-side process of the key job family with a random automated guided vehicle, if the transportation stage of the A-side process of the key job family is selected, replace the automated guided vehicle allocated by the transportation stage of the A-side process of the key job family with a random automated guided vehicle;

[0028] automated guided vehicle exchange, randomly divide the automated guided vehicles allocated by the transportation stages of the B-side process and the A-side process of the key job families and the key job families into two sets, respectively, automated guided vehicle set and automated guided vehicle set , automated guided vehicle set As a key component family The automated guided vehicles (AGVs) sets allocated to the transportation phases of the B-side and A-side processes will be used to... As a key component family The set of automated guided vehicles assigned to the transportation phases of the B-side and A-side processes;

[0029] The execution order of the inner loop neighborhood structure operation is as follows: workpiece exchange of critical workpiece families, workpiece insertion of critical workpiece families, exchange of critical and non-critical workpiece families, exchange of critical workpiece families, mutation of processing units, exchange of processing units, exchange of automated guided vehicles (AGVs), and mutation of AGVs.

[0030] Furthermore, the cooperative search based on sequential intersection and memory mechanism includes cooperative operations based on sequential intersection and cooperative operations based on memory mechanism. The sequential intersection-based cooperative operation generates the workpiece family sequence of the B-face process and the A-face process in the new cooperative solution. The memory mechanism-based cooperative operation generates the workpiece sequence within the workpiece family in the new cooperative solution. The specific process is as follows:

[0031] The binary tournament method is used to select two solutions from the population, and the solution with the smaller objective value is selected as the better solution in the population.

[0032] Perform cooperative operations based on sequential crossover and define the scheduled set. and unscheduled set Used to store the sequence of artifact families, including the scheduled set and unscheduled set Set to empty. Randomly select one process from the B-side process and the A-side process. If the B-side process is selected, randomly select two points in the workpiece family sequence of the B-side process of the population solution to cut in. If the A-side process is selected, randomly select two points in the workpiece family sequence of the A-side process of the population solution to cut in. Divide the workpiece family sequence into the first segment, the middle segment, and the last segment. Fill the workpiece family in the middle segment into the scheduled set in sequence. In the process, the complete sequence of artifact families in the better solution of the population that does not exist in the already scheduled set will be included. The workpiece families are sequentially filled into the unscheduled set. In the middle, the unscheduled set The workpiece family in the sequence replaces the first and last segments of the solution workpiece family sequence of the population in turn. The replaced first and last segments are combined with the middle segments of the solution workpiece family sequence of the population to form the workpiece family sequence of the cooperative new solution.

[0033] Perform memory-based cooperative operations, defining a scheduled set for each artifact family. and unscheduled set Used to store workpieces within a workpiece family, including the scheduled set. Set to null, and assign the sequence of workpieces within the workpiece family to the unscheduled set. For the first workpiece family u When selecting a workpiece, a binary tournament method is used to reselect a population-optimal solution. When the i-th workpiece in the population-optimal solution is... u The workpiece does not exist in the scheduled set. In the middle, the first workpiece family in the population with better solutions will be... u The workpiece is the first workpiece family in the new collaborative solution. u Each workpiece will be part of the unscheduled set. Delete and store in the scheduled set In the middle, when the population has a relatively optimal solution for the workpiece family, the first... u One workpiece exists in the scheduled set. In the unscheduled set Randomly select a workpiece from a workpiece family as the first workpiece in the collaborative new solution. u Each workpiece will be part of the unscheduled set. Delete and store in the scheduled set In the unscheduled set When empty, the workpiece sequence within the workpiece family is completed and forms the workpiece sequence within the workpiece family of the new collaborative solution.

[0034] Furthermore, the specific process of the outer loop variable neighborhood descent search, which embeds an inner loop variable neighborhood descent search and a cooperative search based on sequential crossover and memory mechanisms, is as follows: Traverse the population, starting with the solution of the first population, and execute the outer loop variable neighborhood descent search. The solution of the population is perturbed using the structure of the first outer loop neighborhood to generate a new outer loop solution. This new outer loop solution is decoded using a semi-automatic decoding scheme. The target values ​​of the new outer loop solution and the population solution are compared. If the target value of the new outer loop solution is less than the target value of the population solution, the population solution is replaced with the new outer loop solution, and the population solution is updated. If the target value of the new outer loop solution is greater than the population solution, the population solution remains unchanged. If the target value of the new outer loop solution is less than the target value of the current optimal solution, the current optimal solution is replaced with the new outer loop solution. This process continues until the number of repeated searches using the outer loop neighborhood structure reaches the maximum number of searches for the outer loop variable neighborhood descent search. OWhen the time is reached, the search of the outer loop variable neighborhood structure is ended, the solution of the population enters the inner loop variable neighborhood descent search, the solution of the population is disturbed from the first inner loop neighborhood structure to generate an inner loop new solution, the inner loop new solution is decoded by using the active decoding scheme, the target value of the inner loop new solution is compared with the target value of the solution of the population, if the target value of the inner loop new solution is less than the target value of the solution of the population, the solution of the population is replaced by the inner loop new solution, the solution of the population is updated, if the target value of the inner loop new solution is greater than the target value of the solution of the population, the solution of the population remains unchanged, if the target value of the inner loop new solution is less than the target value of the current optimal solution, the current optimal solution is replaced by the inner loop new solution, if the solution of the population is successfully updated, the number of repeated searches of the solution of the population by the inner loop neighborhood structure is reset to 0, if the number of repeated searches of the solution of the population by the inner loop neighborhood structure reaches the maximum search number of the inner loop variable neighborhood descent I and the solution of the population is updated, the solution of the population is searched from the first inner loop neighborhood structure again, if the number of repeated searches of the solution of the population by the inner loop neighborhood structure reaches the maximum search number of the inner loop variable neighborhood descent I and the solution of the population is not updated, the solution of the population is searched by the next inner loop neighborhood structure, when the search of the eight inner loop neighborhood structures is completed, the inner loop variable neighborhood descent search is ended, the solution of the population performs the cooperative search based on the sequential crossover and the memory mechanism, the sequence part of the cooperative new solution generated by the cooperative operation based on the sequential crossover is decoded by using the semi-automatic decoding scheme, if the target value of the cooperative new solution is less than the target value of the solution of the population, the solution of the population is replaced by the cooperative new solution, if the target value of the cooperative new solution is less than the target value of the current optimal solution, the current optimal solution is replaced by the cooperative new solution, the sequence part of the cooperative new solution generated by the cooperative operation based on the memory mechanism is decoded by using the semi-automatic decoding scheme, if the target value of the cooperative new solution is less than the target value of the solution of the population, the solution of the population is replaced by the cooperative new solution, if the target value of the cooperative new solution is less than the target value of the current optimal solution, the current optimal solution is replaced, if the solution of the population is successfully replaced, the number of executions of the cooperative search based on the sequential crossover and the memory mechanism is reset to 0, when the number of executions of the cooperative search based on the sequential crossover and the memory mechanism reaches the maximum cooperative number C , the cooperative search based on the sequential crossover and the memory mechanism is ended, the solution of the population returns to the outer loop variable neighborhood descent search, if the current optimal solution is replaced in the inner loop variable neighborhood descent search, the solution of the population re-executes the first outer loop neighborhood structure to be disturbed until the search of the four outer loop neighborhood structures is completed, if the current optimal solution is not replaced in the inner loop variable neighborhood descent search, the solution of the population is disturbed by the second outer loop neighborhood structure until the search of the four outer loop neighborhood structures is completed, the next solution of the population starts to perform the outer loop variable neighborhood descent search until all solutions of the population are traversed.

[0035] Further, in the semi-automatic decoding scheme, the allocation rules adopted by each solution in the allocation of processing units in the B-side process and the A-side process and the allocation of automated guided vehicles in the transportation stage of the B-side process and the A-side process are as follows: for the allocation of processing units in the B-side process and the A-side process, the earliest available time rule is adopted as the allocation criterion, that is, from all available processing units, the processing unit that can start processing earliest is preferentially allocated to process the workpiece family; for the allocation of automated guided vehicles, the automated guided vehicles perform transportation tasks in units of workpiece families, and the earliest arrival time rule is adopted for allocation, that is, from all available automated guided vehicles, the automated guided vehicle with the smallest sum of transportation time and idle time between the current workpiece family position is selected, and in the process of semi-automatic decoding, the allocation of workpiece families to processing units and the allocation of workpiece families to automated guided vehicles constitute the allocation part of the solution;

[0036] In the active decoding scheme, the allocation rules adopted by each solution in the allocation of processing units in the B-side process and the A-side process and the allocation of automated guided vehicles in the transportation stage of the B-side process and the A-side process are as follows: for the allocation of processing units in the B-side process and the A-side process, the workpiece family allocation scheme to processing units in the solution is taken as the allocation criterion; for the allocation of automated guided vehicles, the workpiece family allocation scheme to automated guided vehicles in the solution is taken as the allocation criterion.

[0037] Further, the allocation part of the solution in the population adopts a dynamic encoding strategy, and the implementation process of the dynamic encoding strategy is as follows,

[0038] When each solution in the population enters the variable neighborhood descent search in the outer loop and is decoded by the semi-automatic decoding scheme, the allocation part of the solution is automatically generated and saved according to the earliest available time principle, thereby realizing dynamic encoding.

[0039] Further, when the number of consecutive failures of the outer loop new solution and the inner loop new solution to update the solutions of the population reaches the maximum update failure number of the solutions of the population L , population restart is performed, a positive integer smaller than the population size P is randomly generated, the positive integer is denoted as R , the R new solutions are initialized and generated according to the descending order sorting rule of processing time, and the R new solutions are randomly replaced with the R solutions of the population.

[0040] Further, the total running time constraint of the algorithm is taken as the termination condition of the algorithm, the total running time is set to 30×total number of workpieces×number of machines in a processing unit×number of processing units, and the time unit is millisecond.

[0041] The intelligent scheduling method for production and transportation integration in the integrated manufacturing system provided by the application is a kind of knowledge-driven evolutionary algorithm (KGEA) which adopts dynamic coding, carries out sufficient neighborhood exploration on the solutions of the population through inner loop variable neighborhood descent search and outer loop variable neighborhood descent search to obtain better inner loop new solutions and outer loop new solutions, and continuously mines the solutions with smaller target values in the population in the collaborative search based on sequential crossover and memory mechanism to generate the sequence part of the collaborative new solutions, so as to obtain a high-quality solution. L When the number of consecutive failures of the update of the outer loop new solution and the inner loop new solution to the solutions of the population reaches the maximum update failure number of the solutions of the population BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 The flowchart for the implementation of the application is shown in the figure;

[0043] Figure 2 The schematic diagram of a specific solution of the application is shown in the figure;

[0044] Figure 3 The flowchart for the implementation of the collaborative operation based on sequential crossover of the application is shown in the figure;

[0045] Figure 4 The flowchart for the implementation of the collaborative operation based on the memory mechanism of the application is shown in the figure;

[0046] Figure 5 A violin plot for the confidence interval analysis of the present application compared with four existing algorithms.

[0047] Figure 6 A violin plot for the confidence interval analysis of the present application compared with four existing algorithms. DETAILED DESCRIPTION

[0048] As shown in Figures 1-6 , the intelligent scheduling method for production and transportation integration in the integrated manufacturing system provided by the present application is mainly realized through the following steps.

[0049] Step S1: analyze the characteristics of the reconfigurable distributed flow shop group production and transportation integrated scheduling problem in the double-sided surface mounting process of printed circuit boards, determine the minimum maximum completion time as the solving objective, and initialize the parameters, including the population size P , the maximum search number of the outer loop variable neighborhood descent O , the maximum search number of the inner loop variable neighborhood descent I , the maximum number of cooperation C , and the maximum number of solution update failures of the population L .

[0050] Step S2: initialize the population and the current optimal solution, and generate P solutions of the population and one current optimal solution according to the processing time descending order arrangement rule. Among them, the processing time descending order arrangement rule is as follows,

[0051] Compare the processing times of the workpieces in each workpiece family, arrange the workpieces in descending order according to the processing times to generate the workpiece sequence in the workpiece family, sum the processing times of the workpieces in the workpiece family to obtain the processing time of the workpiece family, and arrange the workpiece families in descending order according to the processing times of the workpiece families to generate the workpiece family sequence.

[0052] The population is represented by , each solution is composed of two parts, represents the th solution in the population, is the sequence part, and is a one-dimensional vector, is a two-dimensional vector, n represents the number of workpiece families, is the workpiece family sequence of the B-side flow, is the workpiece family sequence of the A-side flow, is the workpiece sequence in the workpiece family, is a one-dimensional vector representing the workpiece sequence in the workpiece family ; is the assignment part, and are two-dimensional vectors, represent the processing cell assignment scheme of the workpiece families when processing the B-side process and the A-side process, represent the automated guided vehicle assignment scheme of the workpiece families in the transportation stage of the B-side process and the A-side process. The assignment part of the solution in the population adopts a dynamic coding strategy, that is, when each solution in the population enters the variable neighborhood descent search in the outer loop, the assignment part of the solution is automatically generated and saved according to the earliest available time principle when the solution is decoded by the semi-automatic decoding scheme, thereby realizing dynamic coding. In addition, it should be noted that there are multiple processing cells in the printed circuit board mounting factory, each processing cell contains multiple continuous machines, each processing cell can process both the B-side process of the printed circuit board and the A-side process of the printed circuit board, and the A-side process of the printed circuit board must be processed after the B-side process is completed.

[0053] As shown in Figure 2 , in an embodiment of the present application, a population containing three workpiece families is shown, and the solutions of the B-side process and the A-side process are executed in two processing cells . Specifically, the workpiece family sequence represents that the workpiece families are processed in turn according to the order of when processing the B-side process, the workpiece family sequence represents that the workpiece families are processed in turn according to the order of when processing the A-side process, represents the number of workpieces in the workpiece family and the sorting situation, represents that the workpieces in the workpiece family are processed in turn according to the order of , represents that the workpieces in the workpiece family are processed in turn according to the order of , represents that the workpieces in the workpiece family are processed in turn according to the order of , represents the processing cell assignment scheme of the workpiece families when processing the B-side process and the A-side process, represents the processing cells selected by the workpiece family when processing the B-side process are , represents the processing cells selected by the workpiece family when processing the A-side process are , represents the automated guided vehicle assignment scheme of the workpiece families in the transportation stage of the B-side process and the A-side process, represents the automated guided vehicle assignment scheme of the workpiece families in the transportation stage of the B-side process and the A-side process, The automatic guided vehicles selected by the transportation stage of the B-side process are , The automatic guided vehicles selected by the transportation stage of the A-side process are , .

[0054] Step S3: updating the current optimal solution by the outer loop variable neighborhood descent search, which is embedded with the inner loop variable neighborhood descent search and the collaborative search based on the sequential crossover and memory mechanism. The specific introduction is as follows.

[0055] The outer loop variable neighborhood descent search adopts the outer loop neighborhood structure to traverse and optimize the population, and the operation process of the outer loop neighborhood structure includes:

[0056] Workpiece family exchange, randomly selecting a workpiece family sequence of one process from the B-side process and the A-side process, randomly selecting two different workpiece families from the selected workpiece family sequence, and exchanging the positions of the selected two workpiece families in the workpiece family sequence;

[0057] Workpiece family insertion, randomly selecting a workpiece family sequence of one process from the B-side process and the A-side process, randomly selecting a workpiece family and a position in the workpiece family sequence, and inserting the selected workpiece family into the selected position in the workpiece family sequence;

[0058] Workpiece exchange, randomly selecting a workpiece family, randomly selecting two different workpieces from the workpiece sequence in the workpiece family, and exchanging the positions of the selected two workpieces in the workpiece sequence;

[0059] Workpiece insertion, randomly selecting a workpiece family, randomly selecting a workpiece and a position in the workpiece sequence, and inserting the selected workpiece into the selected position in the workpiece sequence;

[0060] The execution order of the operation of the outer loop neighborhood structure is workpiece family insertion, workpiece family exchange, workpiece insertion, and workpiece exchange.

[0061] The inner loop variable neighborhood descent search adopts the inner loop neighborhood structure based on the critical path to traverse and optimize the population, the critical path is the sequence of the workpiece family that determines the maximum completion time in the production scheduling problem, and the workpiece family included in the critical path is the critical workpiece family, and the operation process of the inner loop neighborhood structure includes:

[0062] Critical workpiece family exchange, randomly selecting a workpiece family sequence of one process from the B-side process and the A-side process, randomly selecting two different critical workpiece families on the critical path from the selected workpiece family sequence, and exchanging the positions of the two critical workpiece families in the workpiece family sequence;

[0063] The critical and non-critical workpiece families are swapped. A workpiece family sequence of a process is randomly selected from the B-side process and the A-side process. From the selected workpiece family sequence, a critical workpiece family on the critical path and a normal workpiece family on a non-critical path are randomly selected. The positions of the selected critical workpiece family and the normal workpiece family in the workpiece family sequence are swapped.

[0064] The critical workpiece family swaps the workpieces by randomly selecting a critical workpiece family on the critical path, randomly selecting two different workpieces from the selected critical workpiece family, and swapping the positions of the two selected workpieces in the workpiece sequence within the critical workpiece family.

[0065] For critical workpiece family insertion, randomly select a critical workpiece family on the critical path, randomly select a workpiece from the selected critical workpiece family and a position in the workpiece sequence, and insert the selected workpiece into the position in the selected workpiece sequence.

[0066] The processing unit is mutated by randomly selecting a critical workpiece family on the critical path and randomly selecting either the B-side process or the A-side process of the critical workpiece family. If the B-side process of the critical workpiece family is selected, the processing unit assigned to the B-side process of the critical workpiece family is replaced with a random processing unit. If the A-side process of the critical workpiece family is selected, the processing unit assigned to the A-side process of the critical workpiece family is replaced with a random processing unit.

[0067] Processing unit exchange, randomly selecting critical workpiece families on the critical path. and key component families The processing units allocated to the B-side process and the A-side process are divided into two sets, namely the processing unit set. and processing unit set assemble processing units As a key component family The set of processing units allocated to the B-side process and the A-side process, and the processing unit set As a key component family The set of processing units allocated to the B-side process and the A-side process;

[0068] The automated guided vehicle (AGV) mutates by randomly selecting a critical workpiece family on the critical path and randomly selecting either the B-side or A-side flow of the critical workpiece family for transportation. If the transportation stage of the B-side flow of the critical workpiece family is selected, the AGV assigned to that stage is replaced with a random AGV. If the transportation stage of the A-side flow of the critical workpiece family is selected, the AGV assigned to that stage is replaced with a random AGV.

[0069] Automatic guided vehicle exchange, randomly exchange the key workpiece families on the critical path and the key workpiece families The automatic guided vehicles assigned to the transportation stage of the B-side process and the A-side process are divided into two sets, namely the automatic guided vehicle set and the automatic guided vehicle set The automatic guided vehicle set is the automatic guided vehicle set assigned to the transportation stage of the B-side process and the A-side process of the key workpiece families The automatic guided vehicle set is the automatic guided vehicle set assigned to the transportation stage of the B-side process and the A-side process of the key workpiece families .

[0070] Among them, the execution order of the inner loop neighborhood structure operation is the key workpiece family exchange, the key workpiece family insertion, the key and non-key workpiece family exchange, the key workpiece family exchange, the processing unit mutation, the processing unit exchange, the automatic guided vehicle exchange, and the automatic guided vehicle mutation.

[0071] The collaborative search based on the order crossover and the memory mechanism includes the collaborative operation based on the order crossover and the collaborative operation based on the memory mechanism, the collaborative new solution is generated through the collaborative operation based on the order crossover, and the workpiece sequence in the collaborative new solution is generated through the collaborative operation based on the memory mechanism, and the specific process is as follows:

[0072] The binary tournament method is used to select two solutions from the population and select the solution with a smaller target value as the population optimal solution;

[0073] The collaborative operation based on the order crossover is executed, and the scheduled set and the unscheduled set are defined to store the workpiece family sequence, the scheduled set and the unscheduled set are empty, one process is randomly selected from the B-side process and the A-side process, if the B-side process is selected, the workpiece family sequence of the B-side process of the solution of the population is randomly selected Two points are cut in, if the A-side process is selected, the workpiece family sequence of the A-side process of the solution of the population is randomly selected Two points are cut in, the workpiece family sequence is divided into a first segment, a middle segment and a tail segment, the workpiece families in the middle segment are sequentially filled into the scheduled set , the workpiece families in the complete workpiece family sequence of the population optimal solution that do not exist in the scheduled set are sequentially filled into the unscheduled set , and the unscheduled set The workpiece family in the sequence replaces the first and last segments of the solution workpiece family sequence of the population in turn. The replaced first and last segments are combined with the middle segments of the solution workpiece family sequence of the population to form the workpiece family sequence of the cooperative new solution.

[0074] Perform memory-based cooperative operations, defining a scheduled set for each artifact family. and unscheduled set Used to store workpieces within a workpiece family, including the scheduled set. Set to null, and assign the sequence of workpieces within the workpiece family to the unscheduled set. For the first workpiece family u When selecting a workpiece, a binary tournament method is used to reselect a population-optimal solution. When the i-th workpiece in the population-optimal solution is... u The workpiece does not exist in the scheduled set. In the middle, the first workpiece family in the population with better solutions will be... u The workpiece is the first workpiece family in the new collaborative solution. u Each workpiece will be part of the unscheduled set. Delete and store in the scheduled set In the middle, when the population has a relatively optimal solution for the workpiece family, the first... u One workpiece exists in the scheduled set. In the unscheduled set Randomly select a workpiece from a workpiece family as the first workpiece in the collaborative new solution. u Each workpiece will be part of the unscheduled set. Delete and store in the scheduled set In the unscheduled set When empty, the workpiece sequence within the workpiece family is completed and forms the workpiece sequence within the workpiece family of the new collaborative solution.

[0075] The implementation process of the outer loop variable neighborhood descent search, which incorporates an inner loop variable neighborhood descent search and a cooperative search based on sequential crossover and memory mechanisms, will be explained in detail below.

[0076] The population is traversed, starting from the solution of the first population, the outer loop variable neighborhood descent search is performed, the solution of the population is disturbed by means of the first outer loop neighborhood structure, the outer loop new solution is generated, the outer loop new solution is decoded by adopting the semi-automatic decoding scheme, the objective values of the outer loop new solution and the solution of the population are compared, if the objective value of the outer loop new solution is less than the objective value of the solution of the population, the solution of the population is replaced by the outer loop new solution, the solution of the population is updated, if the objective value of the outer loop new solution is greater than the solution of the population, the solution of the population remains unchanged, if the objective value of the outer loop new solution is less than the objective value of the current optimal solution, the current optimal solution is replaced by the outer loop new solution, when the number of repeated searches of the solution of the population by means of the outer loop neighborhood structure reaches the maximum search number of the outer loop variable neighborhood descent O , the search of the outer loop neighborhood structure is ended, the solution of the population enters the inner loop variable neighborhood descent search, the solution of the population is disturbed starting from the first inner loop neighborhood structure, the inner loop new solution is generated, the inner loop new solution is decoded by adopting the active decoding scheme, the objective values of the inner loop new solution and the solution of the population are compared, if the objective value of the inner loop new solution is less than the objective value of the solution of the population, the solution of the population is replaced by the inner loop new solution, the solution of the population is updated, if the objective value of the inner loop new solution is greater than the solution of the population, the solution of the population remains unchanged, if the objective value of the inner loop new solution is less than the objective value of the current optimal solution, the current optimal solution is replaced by the inner loop new solution, if the solution of the population is successfully updated, the number of repeated searches of the solution of the population by means of the inner loop neighborhood structure is reset to 0, if the number of repeated searches of the solution of the population by means of the inner loop neighborhood structure reaches the maximum search number of the inner loop variable neighborhood descent I , and the solution of the population is updated, the solution of the population is searched again starting from the first inner loop neighborhood structure, if the number of repeated searches of the solution of the population by means of the inner loop neighborhood structure reaches the maximum search number of the inner loop variable neighborhood descent IIf the population solution is not replaced, the population solution is searched for the next inner loop neighborhood structure, and after eight inner loop neighborhood structures are searched, the inner loop variable neighborhood descent search is ended, the population solution performs a cooperative search based on sequential crossover and memory mechanism, the sequence part of the cooperative new solution generated by the cooperative operation based on sequential crossover is decoded by using the semi-automatic decoding scheme, if the target value of the cooperative new solution is less than the target value of the population solution, the cooperative new solution replaces the population solution, if the target value of the cooperative new solution is less than the target value of the current optimal solution, the cooperative new solution replaces the current optimal solution, the sequence part of the cooperative new solution generated by the cooperative operation based on the memory mechanism is decoded by using the semi-automatic decoding scheme, if the target value of the cooperative new solution is less than the target value of the population solution, the cooperative new solution replaces the population solution, if the target value of the cooperative new solution is less than the target value of the current optimal solution, the current optimal solution is replaced, if the population solution is replaced successfully, the execution times of the cooperative search based on sequential crossover and memory mechanism are reset to 0, when the execution times of the cooperative search based on sequential crossover and memory mechanism reach the maximum cooperative times C , the cooperative search based on sequential crossover and memory mechanism is ended, and the population solution returns to the outer loop variable neighborhood descent search, if the current optimal solution is replaced in the inner loop variable neighborhood descent search, the population solution re-executes the first outer loop neighborhood structure for disturbance until four outer loop neighborhood structure searches are completed, if the current optimal solution is not replaced in the inner loop variable neighborhood descent search, the population solution is disturbed by the second outer loop neighborhood structure until four outer loop neighborhood structure searches are completed, and the next population solution starts to execute the outer loop variable neighborhood descent search until all population solutions are traversed.

[0077] In addition, the specific descriptions of the semi-automatic decoding scheme and the active decoding scheme are as follows:

[0078] In the semi-automatic decoding scheme, the allocation rules of the allocation of processing units in the B-side process and the A-side process and the allocation of automatic guided vehicles in the transportation stages of the B-side process and the A-side process are as follows, for the allocation of processing units in the B-side process and the A-side process, the earliest available time rule is used as the allocation criterion, that is, from all available processing units, the processing unit that can start processing earliest is preferentially allocated to process the workpiece family, for the allocation of automatic guided vehicles, the automatic guided vehicles perform transportation tasks in units of workpiece families, and the earliest arrival time rule is used for allocation, that is, from all available automatic guided vehicles, the automatic guided vehicle with the smallest sum of transportation time and idle time between the current workpiece family position is selected, in the process of semi-automatic decoding, the allocation of workpiece families to processing units and the allocation of workpiece families to automatic guided vehicles constitute the allocation part of the solution;

[0079] In the active decoding scheme, the allocation rule adopted by each solution in the allocation of processing units in the B-side process and the A-side process and the allocation of automated guided vehicles in the transportation stage of the B-side process and the A-side process is as follows: for the allocation of processing units in the B-side process and the A-side process, the solution adopts the workpiece family allocation scheme in the solution as the allocation criterion; for the allocation of automated guided vehicles, the automated guided vehicle executes the transportation task in the unit of workpiece family, and the solution adopts the workpiece family allocation scheme in the solution as the allocation criterion.

[0080] By Figure 3 the implementation flowchart of the cooperative operation based on the sequential crossover of the application shown in FIG. Figure 4 the implementation flowchart of the cooperative operation based on the memory mechanism of the application shown in FIG. 5, further shows the process of the sequential part of the cooperative new solution generated by the cooperative search based on the sequential crossover and the memory mechanism from the five solutions. Figure 3 and Figure 4 ①②③④⑤ in the above table represent the serial numbers of the solutions, the solution ① represents the solution executing the cooperative search based on the sequential crossover and the memory mechanism, the solutions ②③④⑤ are four solutions in the randomly selected population, and the five solutions are illustrated by taking the workpiece family sequence of the three workpiece families in the B-side process and the A-side process and the workpiece sequence in the workpiece family as examples. Figure 3 The cooperative operation based on the sequential crossover is shown in the following steps. The scheduled set of the workpiece families is set as , the unscheduled set of the workpiece families is set as , the scheduled set of the workpiece families is set as empty, and the unscheduled set of the workpiece families is filled with the three workpiece families in sequence. For the cooperative operation of the workpiece family sequence in the B-side process, the solution ② and the solution ③ are randomly selected from the population, the target value of the solution ② and the target value of the solution ③ are compared, the solution ③ with the smaller target value is selected, the workpiece family sequence selection segment 1 of the B-side process of the solution ① is cut, and the workpiece families in the segment are filled into the scheduled set of the workpiece families , the workpiece families in the segment 1 are deleted from the unscheduled set of the workpiece families , the workpiece families in the workpiece family sequence of the B-side process of the solution ③ are traversed, if the traversed workpiece family exists in the unscheduled set of the workpiece families , the workpiece family is sequentially filled into the segment 2, the workpiece family is deleted from the unscheduled set of the workpiece families and filled into the scheduled set of the workpiece families , and when the unscheduled set of the workpiece families When empty, it indicates that a new B-side process workpiece family sequence has been generated by the cooperative operation based on sequential intersection. The new B-side process workpiece family sequence, together with the A-side process workpiece family sequence and the workpiece sequences within the workpiece family in solution ①, constitutes the sequence part of the new cooperative solution 1. When cooperating on the A-side process workpiece family sequence, solutions ④ and ⑤ are randomly selected from the population. The objective values ​​of solutions ④ and ⑤ are compared, and solution ⑤ with the smaller objective value is selected. The A-side process workpiece family sequence of solution ① is segmented into fragment 1, and the workpiece families in the fragment are filled into the scheduled set. In the middle, the workpiece family in fragment 1 is never scheduled set Delete the workpiece family sequence of the A-side process of solution ⑤, and if the traversed workpiece family exists in the unscheduled set. In the middle, the workpiece family will be filled into fragment 2 in sequence, and the workpiece family will be placed in the unscheduled set. Delete and populate the scheduled set When the unscheduled set When empty, it indicates that the cooperative operation based on sequential intersection has generated a new workpiece family sequence of A-face process. The new workpiece family sequence of A-face process, together with the workpiece family sequence of B-face process in solution ① and the workpiece sequence within the workpiece family, constitutes the sequence part of the new cooperative solution 2.

[0081] Figure 4 Demonstrates memory-based cooperative operations for workpiece families. When coordinating workpiece sequences within a given timeframe, set the already scheduled set of workpieces. and unscheduled set The already scheduled set Set to empty, and set the workpiece family The workpieces within are sequentially filled into the unscheduled set. Generate a new collaborative solution for the workpiece family. When the first workpiece is selected, the binary tournament rules choose solution ②, and the workpiece family in solution ② is... The first workpiece Not present in the scheduled set Then the workpiece family in solution ② medium workpiece As a family of workpieces in the new collaborative solution three The first workpiece in the solution will collaborate with the workpiece family of the new solution three. The first workpiece in the unscheduled set Delete and store in the scheduled set Generate a new collaborative solution for the workpiece family. When selecting the second workpiece, the binary tournament rules choose solution ②, and the workpiece family in solution ②... The second workpiece Not present in the scheduled set , the first job in the job family of the cooperative new solution three, the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the first job in the job family of the cooperative new solution three , the second job in the job family of the cooperative new solution three, the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set , the second job in the job family of the cooperative new solution three is deleted from the unscheduled set and stored in the scheduled set Not present in the scheduled set Then the workpiece family in solution ④ medium workpiece As a family of workpieces in the new collaborative solution three The first workpiece in the solution will collaborate with the workpiece family of the new solution three. The first workpiece in the unscheduled set Delete and store in the scheduled set Generate a new collaborative solution for the workpiece family. When choosing the second workpiece in the process, the binary tournament rules select solution ④, which is the second workpiece in the workpiece family of solution ④. Not present in the scheduled set Then the workpiece family in solution ④ medium workpiece As a family of workpieces in the new collaborative solution three The second workpiece in the solution will collaborate with the workpiece family of the new solution three. The second workpiece in the unscheduled set Delete and store in the scheduled set Generate a new collaborative solution for the workpiece family. When the third workpiece is selected, the binary tournament rules choose solution ⑤, and the workpiece family in solution ⑤ is... The third workpiece Existing in the scheduled set Then the never-scheduled set Randomly select a workpiece family The workpieces in the solution are considered as a family of workpieces in the new collaborative solution 3. The third workpiece in the solution will collaborate with the new solution's workpiece family. The third workpiece in the never-scheduled set Delete and store in the scheduled set Generate a new collaborative solution for the workpiece family. When dealing with the fourth workpiece, the binary tournament rules select solution ⑤, and the workpiece family in solution ⑤... The fourth workpiece Existing in the scheduled set Then the never-scheduled set Randomly select a workpiece family The workpieces in the solution are considered as a family of workpieces in the new collaborative solution 3. The fourth workpiece in the solution will collaborate with the workpiece family of the new solution three. The fourth workpiece in the unscheduled set Delete and store in the scheduled set The complete sequence of the workpiece family in the new collaborative solution 3 is generated and together with the workpiece family sequence of solution ① constitutes the sequence part of the new collaborative solution 3.

[0082] Step S4: If the number of consecutive failures of the outer loop variable neighborhood descent search and the inner loop variable neighborhood descent search in updating the solution of the population reaches the maximum number of solution update failures of the population. L Perform a population restart and randomly generate a population smaller than the population size. P A positive integer, denoted as R Initialize and generate according to the descending order of processing time. R A new solution, and will R A new solution randomly replaces the population. R One solution.

[0083] Step S5: Determine if the termination condition has been met. If it has, the algorithm ends and the current optimal solution is output; otherwise, return to step S3. The total running time constraint is used as the termination condition, and the total running time is set to 30. Total number of workpieces Number of machines in the processing unit Number of processing units, time unit is milliseconds.

[0084] To better illustrate the effectiveness of this invention in solving the problem of integrated scheduling of group production and transportation in reconfigurable distributed flow workshops, a series of experimental analyses of examples of this invention are provided to further describe and illustrate the invention.

[0085] In this invention, the problem size is determined by the number of workpiece families. Number of processing units Number of machines in the processing unit and the number of automated guided vehicles Four factors were combined to determine 36 different problem-scale configurations. For each problem-scale configuration, 5 test cases were randomly generated, resulting in a total of 180 cases. The number of workpieces within each workpiece family was distributed in the range of [3,5]. The processing time for each workpiece on different machines was evenly distributed in the range of [50,100]. The setup time for processing units before processing different workpiece families was evenly distributed in the range of [100,150] to simulate uncertainties in actual production. In the distributed assembly line workshop, warehouses and processing units are arranged in a matrix. x The axis represents the row. y The axes represent columns. Using the coordinate system described above, the positions of warehouses and processing units in the distributed assembly line layout can be determined. The Manhattan distance formula is used to calculate the distances between the warehouse and processing units, and between processing units. The transport time for automated guided vehicles (AGVs) to transport workpiece families is determined by the transport distance. The Manhattan distance formula is calculated as follows.

[0086]

[0087] Represents warehouse distance between the processing units to the processing units , represents the row in which the warehouse is located, represents the column in which the warehouse is located, represents the row in which the processing unit is located, represents the column in which the processing unit is located;

[0088]

[0089] represents the row in which the processing unit is located, distance between the processing units to the processing units , represents the row in which the processing unit is located, represents the column in which the processing unit is located; represents the row in which the processing unit is located, represents the column in which the processing unit is located;

[0090] The termination condition of the KGEA algorithm of the application is the total running time of the algorithm, the total running time is set to 30xtotal number of workpiecesxnumber of machines in the processing unitxnumber of processing units, the time unit is millisecond, for each instance, independently run 10 times, so the relative percentage increase value (RPI) of each instance can be obtained. Then, group the instances under the same scale to obtain the average relative percentage increase value (ARPI) of each combination.

[0091] In terms of parameter setting, in order to better solve and optimize the reconfigurable distributed flow shop group production and transportation integrated scheduling problem, the population size P is set to 10, the maximum search number in the outer loop neighborhood search O is set to 3, the maximum search number in the inner loop neighborhood search I is set to 10, the maximum number of cooperation C is set to 15, the maximum number of solution update failures of the population L is set to 79.

[0092] ​​​​​​​​The experimental results and analysis of the present example are as follows. After the KGEA algorithm of the present application is completed, parameter setting is performed, and systematic experiments are performed with four comparison algorithms. The four existing algorithms include a nested variable neighborhood descent algorithm (NVND), a greedy iteration algorithm based on population cooperation (CPIG), a two-stage greedy iteration algorithm (TIG), and a new cooperative artificial bee colony algorithm (NCABC). To ensure fairness of comparison, all algorithms use a uniform problem size, adopt the same objective, and introduce the same workpiece family allocation rule and automated guided vehicle allocation rule. Finally, the solving quality and stability of the KGEA of the present application are evaluated according to the average relative percentage increase value obtained by each algorithm under the same problem size. The example comparison results of the present application and the four existing algorithms are shown in Table 1.

[0093] Table 1. Example comparison results of the present application and the four existing algorithms

[0094]

[0095] As can be seen from Table 1, the ARPI value obtained by KGEA is the smallest in most problem sizes. For example, in the 36th problem size 20-8-8-7, the ARPI value of KGEA is 1.065, while the ARPI values of the other algorithms, i.e., CPIG (4.14), NVND (5.468), TIG (32.423), and NCABC (12.302), are significantly larger. As can be seen from the average values of APRI obtained by the five algorithms under different sizes, the average value of ARPI of KGEA is 1.38, which is significantly lower than the average values of ARPI of the other algorithms, i.e., CPIG (3.07), NVND (5.08), TIG (31.01), and NCABC (9.86). Thus, it is shown that KGEA is superior to the four existing comparison algorithms in overall performance. In summary, the KGEA algorithm of the present application has shown significant advantages in most problem sizes, and can obtain better optimization results when solving complex production scheduling problems. This result fully verifies the advantages of the KGEA algorithm of the present application in solving performance. Figure 5 An example algorithm confidence interval analysis graph is plotted based on the ARPI data obtained by comparison of the existing algorithms in Table 1. From Figure 5 As can be seen, the confidence interval of KGEA is significantly narrower than that of the other four existing algorithms compared with the present application, and there is no overlap with the confidence intervals of the other four existing algorithms compared with the present application. This further indicates that the KGEA algorithm of the present application has smaller dispersion and higher stability, and further verifies the significant advantages of the KGEA of the present application in data stability and concentration. Figure 6For the example algorithm violin analysis chart drawn based on the ARPI data obtained by comparing the existing algorithms in table 1, the data set corresponding to KGEA is concentrated in a narrow range near zero, which indicates that its dispersion is extremely low and its performance is extremely stable, and these comparisons show that the KGEA algorithm of the present application performs best in data set and dispersion control, and the stability of its results is obviously better than that of the four existing algorithms compared, proving that the present application is a reliable method in terms of solution consistency.

[0096] In summary, the present application can effectively solve the reconfigurable distributed flow shop group production and transportation integrated scheduling problem. The present application is closer to the actual industrial production environment, and comprehensively considers the coordination relationship between production and logistics. By jointly optimizing the workpiece family sequence, workpiece sequence, machining unit allocation and automated guided vehicle allocation, the deep coordination of the production and transportation process is realized, and the overall scheduling efficiency and resource utilization are improved.

Claims

1. An intelligent scheduling method for integrating production and transportation in an integrated manufacturing system, characterized in that, Includes the following steps, S1. Analyze the characteristics of the reconfigurable distributed flow shop group production and transportation integration scheduling problem in the double-sided mounting process of printed circuit boards, determine that minimizing the maximum completion time is the solution objective, and initialize parameters, including population size. P The outer loop changes the neighborhood, resulting in a decrease in the maximum number of searches. O The maximum number of searches decreased due to the change in the inner loop neighborhood. I Maximum number of collaborations C The maximum number of update failures for a population L ; S2. Initialize the population and the current optimal solution, and generate the population according to the descending order of processing time. P There are one solution and one current optimal solution, where the population consists of... This indicates that each solution It consists of two parts. The first digit of the population There are several solutions, and each solution contains a family of workpieces that require sequential processing of the B-side and A-side of the printed circuit board. For the sequence part, and It is a one-dimensional vector. It is a two-dimensional vector. n Indicates the number of workpiece families. This represents the sequence of workpiece families for processing surface B. For the workpiece family sequence of the A-side processing flow, For a sequence of workpieces within a workpiece family, A one-dimensional vector representing a family of workpieces. The sequence of workpieces within; For the allocation portion, and Both are two-dimensional vectors. This indicates the machining unit allocation scheme selected by the workpiece family when executing the B-side and A-side processes. This represents the automated guided vehicle (AGV) allocation scheme for the transportation stages of the B-side and A-side processes of a workpiece family. S3. Update the current optimal solution through an outer loop variable neighborhood descent search. This outer loop variable neighborhood descent search embeds an inner loop variable neighborhood descent search and a cooperative search based on sequential crossover and memory mechanisms. The cooperative search based on sequential crossover and memory mechanisms includes cooperative operations based on sequential crossover and cooperative operations based on memory mechanisms. The cooperative operations based on sequential crossover generate the workpiece family sequences for the B-face and A-face processes in the new cooperative solution. The cooperative operations based on memory mechanisms generate the workpiece sequences within the workpiece families in the new cooperative solution. The specific process is as follows. The binary tournament method is used to select two solutions from the population, and the solution with the smaller objective value is selected as the better solution in the population. Perform cooperative operations based on sequential crossover and define the scheduled set. and unscheduled set Used to store the sequence of artifact families, including the scheduled set and unscheduled set Set to empty. Randomly select one process from the B-side process and the A-side process. If the B-side process is selected, randomly select two points in the workpiece family sequence of the B-side process of the population solution to cut in. If the A-side process is selected, randomly select two points in the workpiece family sequence of the A-side process of the population solution to cut in. Divide the workpiece family sequence into the first segment, the middle segment, and the last segment. Fill the workpiece family in the middle segment into the scheduled set in sequence. In the process, the complete sequence of artifact families in the better solution of the population that does not exist in the already scheduled set will be included. The workpiece families are sequentially filled into the unscheduled set. In the middle, the unscheduled set The workpiece family in the sequence replaces the first and last segments of the solution workpiece family sequence of the population in turn. The replaced first and last segments are combined with the middle segments of the solution workpiece family sequence of the population to form the workpiece family sequence of the cooperative new solution. Perform memory-based cooperative operations, defining a scheduled set for each artifact family. and unscheduled set Used to store workpieces within a workpiece family, including the scheduled set. Set to null, and assign the sequence of workpieces within the workpiece family to the unscheduled set. For the first workpiece family u When selecting a workpiece, a binary tournament method is used to reselect a population-optimal solution. When the i-th workpiece in the population-optimal solution is... u The workpiece does not exist in the scheduled set. In the middle, the first workpiece family in the population with better solutions will be... u The workpiece is the first workpiece family in the new collaborative solution. u Each workpiece will be part of the unscheduled set. Delete and store in the scheduled set In the middle, when the population has a relatively optimal solution for the workpiece family, the first... u One workpiece exists in the scheduled set. In the unscheduled set Randomly select a workpiece from a workpiece family as the first workpiece in the collaborative new solution. u Each workpiece will be part of the unscheduled set. Delete and store in the scheduled set In the unscheduled set When empty, the workpiece sequence within the workpiece family is completed and forms the workpiece sequence within the workpiece family of the new collaborative solution; S4. If the number of consecutive failures in updating the solution of the population by the outer loop variable neighborhood descent search and the inner loop variable neighborhood descent search reaches the maximum number of solution update failures of the population. L Perform population restart; S5. Determine if the termination condition has been met. If it has, the algorithm ends and the current optimal solution is output. Otherwise, return to S3 to continue the search.

2. The intelligent scheduling method for integrated production and transportation in an integrated manufacturing system according to claim 1, characterized in that, The solutions of the population and the sequence of the current best solution are generated according to the descending order of processing time, as follows. The processing times of the workpieces within each workpiece family are compared, and the workpieces are sorted in descending order according to their processing times to generate a workpiece sequence within the workpiece family. The processing times of the workpieces within the workpiece family are summed and recorded as the processing time of the workpiece family. The workpiece families are then sorted in descending order according to their processing times to generate a workpiece family sequence.

3. The intelligent scheduling method for integrated production and transportation in an integrated manufacturing system according to claim 2, characterized in that, The outer-loop variable neighborhood descent search optimizes population traversal using an outer-loop neighborhood structure. The operation process of the outer-loop neighborhood structure includes... Workpiece family exchange: Randomly select a workpiece family sequence from the B-side process and the A-side process, randomly select two different workpiece families from the selected workpiece family sequence, and exchange the positions of the two selected workpiece families in the workpiece family sequence. Workpiece family insertion: Randomly select a workpiece family sequence from the B-side process and the A-side process, randomly select a workpiece family and a position from the selected workpiece family sequence, and insert the selected workpiece family into the selected position in the workpiece family sequence. Workpiece swapping involves randomly selecting a workpiece family, randomly selecting two different workpieces from the workpiece sequence within that workpiece family, and swapping the positions of the two selected workpieces in the workpiece sequence. Workpiece insertion: Randomly select a workpiece family, randomly select a workpiece and a position from the workpiece sequence within the workpiece family, and insert the selected workpiece into the selected position in the workpiece sequence; The execution order of operations in the outer loop neighborhood structure is: workpiece family insertion, workpiece family swap, workpiece insertion, and workpiece swap.

4. The intelligent scheduling method for integrated production and transportation in the integrated manufacturing system according to claim 3, characterized in that, The inner-loop variable neighborhood descent search employs an inner-loop neighborhood structure based on the critical path to traverse and optimize the population. The critical path refers to the sequence of job families that determines the maximum completion time in the production scheduling problem; the job families contained in the critical path are the critical job families. The operation process of the inner-loop neighborhood structure includes... Critical workpiece family swapping: Randomly select a workpiece family sequence from the B-side process and the A-side process, randomly select two different critical workpiece families on the critical path from the selected workpiece family sequence, and swap the positions of these two critical workpiece families in the workpiece family sequence. The critical and non-critical workpiece families are swapped. A workpiece family sequence of a process is randomly selected from the B-side process and the A-side process. From the selected workpiece family sequence, a critical workpiece family on the critical path and a normal workpiece family on a non-critical path are randomly selected. The positions of the selected critical workpiece family and the normal workpiece family in the workpiece family sequence are swapped. The critical workpiece family swaps the workpieces by randomly selecting a critical workpiece family on the critical path, randomly selecting two different workpieces from the selected critical workpiece family, and swapping the positions of the two selected workpieces in the workpiece sequence within the critical workpiece family. For critical workpiece family insertion, randomly select a critical workpiece family on the critical path, randomly select a workpiece from the selected critical workpiece family and a position in the workpiece sequence, and insert the selected workpiece into the position in the selected workpiece sequence. The processing unit is mutated by randomly selecting a critical workpiece family on the critical path and randomly selecting either the B-side process or the A-side process of the critical workpiece family. If the B-side process of the critical workpiece family is selected, the processing unit assigned to the B-side process of the critical workpiece family is replaced with a random processing unit. If the A-side process of the critical workpiece family is selected, the processing unit assigned to the A-side process of the critical workpiece family is replaced with a random processing unit. Processing unit exchange, randomly selecting critical workpiece families on the critical path. and key component families The processing units allocated to the B-side process and the A-side process are divided into two sets, namely the processing unit set. and processing unit set assemble processing units As a key component family The set of processing units allocated to the B-side process and the A-side process, and the processing unit set As a key component family The set of processing units allocated to the B-side process and the A-side process; The automated guided vehicle (AGV) mutates by randomly selecting a critical workpiece family on the critical path and randomly selecting either the B-side or A-side flow of the critical workpiece family for transportation. If the transportation stage of the B-side flow of the critical workpiece family is selected, the AGV assigned to that stage is replaced with a random AGV. If the transportation stage of the A-side flow of the critical workpiece family is selected, the AGV assigned to that stage is replaced with a random AGV. Automated Guided Vehicle (AGV) swapping, randomly selecting critical workpiece families on the critical path. and key component families The automated guided vehicles (AGVs) allocated to the transportation phases of both the B-side and A-side processes are divided into two sets: the AGV set and the AGV set. and automated guided vehicles assemble automated guided vehicles As a key component family The automated guided vehicles (AGVs) sets allocated to the transportation phases of the B-side and A-side processes will be used to... As a key component family The set of automated guided vehicles assigned to the transportation phases of the B-side and A-side processes; The execution order of the inner loop neighborhood structure operation is as follows: workpiece exchange of critical workpiece families, workpiece insertion of critical workpiece families, exchange of critical and non-critical workpiece families, exchange of critical workpiece families, mutation of processing units, exchange of processing units, exchange of automated guided vehicles (AGVs), and mutation of AGVs.

5. The intelligent scheduling method for integrated production and transportation in an integrated manufacturing system according to claim 4, characterized in that, The specific process of the outer loop variable neighborhood descent search, which embeds an inner loop variable neighborhood descent search and a cooperative search based on sequential crossover and memory mechanisms, is as follows: Traverse the population, starting with the solution of the first population, and perform an outer loop variable neighborhood descent search. The solution of the population is perturbed using the structure of the first outer loop neighborhood to generate a new outer loop solution. This new outer loop solution is decoded using a semi-automatic decoding scheme. The target values ​​of the new outer loop solution and the population solution are compared. If the target value of the new outer loop solution is less than the target value of the population solution, the population solution is replaced with the new outer loop solution, and the population solution is updated. If the target value of the new outer loop solution is greater than the population solution, the population solution remains unchanged. If the target value of the new outer loop solution is less than the target value of the current optimal solution, the current optimal solution is replaced with the new outer loop solution. This process continues until the number of repeated searches using the outer loop neighborhood structure reaches the maximum number of searches for the outer loop variable neighborhood descent search. O When the search of the outer loop neighborhood structure ends, the population's solution enters the inner loop variable neighborhood descent search. Starting from the first inner loop neighborhood structure, the population's solution is perturbed to generate a new inner loop solution. The new inner loop solution is decoded using an active decoding scheme. The target values ​​of the new inner loop solution and the population solution are compared. If the target value of the new inner loop solution is less than the target value of the population solution, the population solution is replaced with the new inner loop solution, and the population solution is updated. If the target value of the new inner loop solution is greater than the population solution, the population solution remains unchanged. If the target value of the new inner loop solution is less than the target value of the current best solution, the current best solution is replaced with the new inner loop solution. If the population solution is successfully updated, the number of times the inner loop neighborhood structure performs repeated searches on the population's solution is reset to 0. If the number of repeated searches on the population's solution using the inner loop neighborhood structure reaches the maximum number of searches in the inner loop variable neighborhood descent search... I If the population's solution is updated, the search for the population's solution restarts from the first inner loop neighborhood structure. If the number of times the population's solution is repeatedly searched using the inner loop neighborhood structure reaches the maximum number of searches required for inner loop variable neighborhood descent, then... I If the population's solution is not updated, the next inner loop neighborhood structure search is performed on the population's solution. After the search of eight inner loop neighborhood structures is completed, the inner loop variable neighborhood descent search ends, and the population's solution undergoes a cooperative search based on sequential crossover and memory mechanisms. For the sequence portion of the cooperative new solution generated by the sequential crossover cooperative operation, a semi-automatic decoding scheme is used for decoding. If the target value of the cooperative new solution is less than the target value of the population's solution, the cooperative new solution replaces the population's solution. If the target value of the cooperative new solution is less than the target value of the current best solution, the cooperative new solution replaces the current best solution. For the sequence portion of the cooperative new solution generated by the memory mechanism cooperative operation, a semi-automatic decoding scheme is used for decoding. If the target value of the cooperative new solution is less than the target value of the population's solution, the cooperative new solution replaces the population's solution. If the target value of the cooperative new solution is less than the target value of the current best solution, the current best solution is replaced. If the population's solution is successfully replaced, the execution count of the cooperative search based on sequential crossover and memory mechanisms is reset to 0. When the execution count of the cooperative search based on sequential crossover and memory mechanisms reaches the maximum number of cooperative searches... C Afterwards, the cooperative search based on sequential crossover and memory mechanisms ends, and the solutions of the population return to the outer loop variable neighborhood descent search. If the current optimal solution is replaced in the inner loop variable neighborhood descent search, the solutions of the population re-execute the first outer loop neighborhood structure for perturbation until the search of four outer loop neighborhood structures is completed. If the current optimal solution is not replaced in the inner loop variable neighborhood descent search, the solutions of the population are perturbed through the second outer loop neighborhood structure until the search of four outer loop neighborhood structures is completed. The solutions of the next population begin to execute the outer loop variable neighborhood descent search until all population solutions are traversed.

6. The intelligent scheduling method for integrated production and transportation in an integrated manufacturing system according to claim 5, characterized in that, In the semi-automatic decoding scheme, the allocation rules for each solution's processing units in the B-side and A-side processes, as well as the allocation of automated guided vehicles (AGVs) in the transportation phases of the B-side and A-side processes, are as follows: For the allocation of processing units in the B-side and A-side processes, the earliest available time rule is used as the allocation criterion, that is, from all currently available processing units, the processing unit that can start processing earliest is prioritized to process the workpiece family; for the allocation of AGVs, AGVs perform transportation tasks on a workpiece family basis, and the earliest arrival time rule is used for allocation, that is, from all available AGVs, the AGV with the smallest sum of transportation time and idle time between the current workpiece family's location and its location is selected. In the semi-automatic decoding process, the allocation of workpiece families to processing units and the allocation of workpiece families to AGVs constitute the solution allocation part. In the active decoding scheme, the allocation rules used for each solution when allocating processing units in the B-side and A-side processes, and when allocating automated guided vehicles (AGVs) in the transportation phase of the B-side and A-side processes, are as follows: For the allocation of processing units in the B-side and A-side processes, the workpiece family allocation scheme in the solution is used as the allocation criterion; for the allocation of AGVs, AGVs perform transportation tasks on a workpiece family basis, and the workpiece family allocation scheme in the solution is used as the allocation criterion.

7. The intelligent scheduling method for integrated production and transportation in an integrated manufacturing system according to claim 6, characterized in that, The solution allocation for the population employs a dynamic encoding strategy, the implementation process of which is as follows. When each solution in the population enters the outer loop variable neighborhood descent search and is decoded through a semi-automatic decoding scheme, the solution allocation part is automatically generated and saved according to the earliest available time principle, thereby realizing dynamic encoding.

8. The intelligent scheduling method for integrated production and transportation in an integrated manufacturing system according to claim 3, characterized in that, When the number of consecutive failures in updating the population's solution using new solutions from the outer and inner loops reaches the maximum number of solution update failures for the population. L When necessary, the population is restarted, and a new population smaller than the original population size is randomly generated. P A positive integer, denoted as R Initialize and generate according to the descending order of processing time. R A new solution, and will R A new solution randomly replaces the population. R One solution.

Images