A flexible job shop scheduling method considering limited processing machine resources and multiple AGVs

Abstract

A flexible job shop scheduling method considering limited processing machine resources and multiple AGVs includes the following steps: S1, setting problem data and algorithm parameters; S2, generating an initial population using three-segment coding; S3, evaluating each chromosome using an improved decoding strategy; S4, judging whether the maximum number of algorithm iterations I has been reached, if so, jumping to S11; otherwise, s=s+1, jumping to S5; S5, calculating the fitness value and sorting the population; S6, selecting (m+1) chromosomes through the dual strategies of elite retention and tournament selection; S7, alternately selecting groups; S8, selecting parent chromosomes P1 and P2 from sets X1 and X2 respectively according to crossover probability Pc; S9, selecting parent chromosomes P from the chromosome population according to mutation probability Pm, and obtaining offspring C through random number optimization mutation and selection mutation; S10, updating the population, replacing the initial population X with the mutated population init , and replace the (m+1) chromosomes at the end of the population sorting with the (m+1) chromosomes selected by the dual strategies of elite retention and tournament selection to obtain the evolved new population, and jump to S3; S11, output the optimal factory logistics scheduling plan, calculate the fitness values ​​of all populations, and select the chromosome individual with the smallest fitness value for decoding to obtain the optimal factory logistics scheduling plan.

Description

Technical Field

[0001] The present invention relates to the technical field of production scheduling of discrete manufacturing systems, and in particular to a flexible job shop scheduling method considering limited processing machine resources and multiple AGVs. Background Art

[0002] Currently, flexible job shop scheduling methods primarily assume abundant machine resources, meaning that each workpiece process can be processed on any available machine. For example, Zhang Jianxin proposed a randomized greedy initialization population genetic algorithm for flexible job shop scheduling (Inner Mongolia University of Technology: CN 202210095587.3.X[P].2022-01-26). This algorithm solves the flexible job shop scheduling problem by designing a randomized greedy initialization population genetic algorithm based on the relatively shortest machine processing time. Liu Zhifeng proposed a flexible job shop scheduling method based on a multi-level neighborhood structure and a hybrid genetic algorithm (Beijing University of Technology: CN202011056359.2[P].2024-03-22.). This method improves the accuracy and convergence speed of the genetic algorithm for the flexible job shop scheduling problem by designing an initialization strategy based on hybrid heuristic rules and effectively searching a new multi-level neighborhood structure based on key processes. However, neither of these methods fully considers the scenarios of limited machine resources and multi-AGV logistics scheduling.

[0003] However, in practical flexible job shops, processing machine resources are often limited, and most processes can only be performed on a subset of available machines. Traditional genetic algorithms, when dealing with such scheduling problems, often struggle to match valid processing machines during individual decoding. This results in wasted computing resources in the search space, low optimization efficiency, and slow convergence.

[0004] Furthermore, in modern smart factories, AGVs are used throughout the entire production process, including the transportation, loading and unloading, and storage of raw materials, semi-finished products, and finished products. Currently, scheduling methods for flexible job shops primarily focus on production scheduling design, lacking consideration for multi-AGV logistics scheduling.

[0005] Therefore, the existing flexible job shop scheduling method has poor adaptability in an environment with limited processing machine resources, and it is difficult to come up with an effective scheduling plan. Moreover, the scheduling plan obtained lacks consideration of multi-AGV logistics scheduling compared with actual production. Summary of the Invention

[0006] The present invention aims to solve the problem that the existing flexible job shop scheduling method has poor adaptability in an environment with limited processing machine resources, it is difficult to obtain an effective scheduling plan, and the scheduling plan obtained lacks consideration for multi-AGV logistics scheduling compared with actual production. A flexible job shop scheduling method that takes into account limited processing machine resources and multiple AGVs is provided.

[0007] The present invention simultaneously considers the flexible job shop scheduling problems of limited processing machine resources and multiple AGVs, establishes a mathematical model with the limited processing machine resource constraints, multiple AGV logistics rule constraints, and minimization of the maximum completion time as the goal, introduces a three-segment chromosome encoding strategy into the improved genetic algorithm to simultaneously solve the three sub-problems of process processing sequence, process selected processing machine, and process selected logistics transportation equipment, adopts random number optimization mutation and elite retention, tournament selection dual strategies to generate and retain excellent individual codes, improves the algorithm's search accuracy and convergence speed, and adds a greedy algorithm in decoding to further reduce the minimization of the maximum completion time.

[0008] The flexible job shop scheduling problem considering limited processing machine resources and multiple AGVs in this invention is described as follows:

[0009] Assume there are i workpiece sets {j1,…,j i}, each workpiece contains several processes O ij (i=1,2,…,n;j=1,…,n i ), each process needs to be assigned to any one of the k machines for processing, and the set of k processing machines can be expressed as {M1,…,M k Each process needs to be transported by any one of m AGVs. The set of m AGVs can be expressed as {A1,…,A m}.

[0010] The optimization goal of the algorithm is to minimize the maximum completion time, and the objective function is: Among them E ijkm The jth process of workpiece i on processing machine M k The end time when the previous processing is completed and transported to the next processing machine by the mth AGV.

[0011] In this flexible job shop scheduling problem, there are four constraints: limited processing machine resources, multi-AGV transportation rules, process sequence and workpiece resources.

[0012] The limited processing machine resource constraint means that some processes can only be processed on specific processing machines, which can be expressed as: S' ijk =S ijk +T ijk +L(1-Pijk ), where S′ ijk For process O ij On machine M k The time when the processing is completed; L represents an infinite positive number; P ijk Indicates process O ij Is it possible to process the machine M k If it is possible, it is 1, otherwise it is 0;

[0013] The multi-AGV transportation rule constraint refers to the constraint relationship between the AGV and the workpiece and processing machine during transportation. i1k ≥ST ijm +w ikm The first step of the AGV transporting workpiece i is to transport it from the warehouse loading area to the processing machine M. k Department, ST ijm Indicates that AGV performs delivery task O ij The start time of transportation, w ikm Indicates that the mth AGV transports workpiece i from the warehouse loading area to the processing machine M k Transportation time; E ink +w′ ikm ≤ET ijm Indicates that the last process of AGV transporting workpiece i must be from processing machine M k Transport to warehouse unloading area, ET ijm Indicates that AGV performs delivery task O ij The end transportation time, w′ ikm Indicates that the mth AGV takes workpiece i from processing machine M k Transportation time to the warehouse loading area; max(ET ijm ,E ijk )≤S ijk +L(1-Y ijk ) indicates that the mth AGV transports workpiece i to processing machine M k When the machine completes the previous processing task, it must wait for the machine to complete the previous processing task and start the processing step O ij You can leave after ijk Indicates processing machine M k Is it possible to process process O? ij , if it can be processed, it is 1, otherwise it is 0;

[0014] The process sequence constraint means that some processes in the workpiece processing process need to be completed before the previous process can be carried out, which can be expressed as: S i(j-1)k +T ijk ≤S ijk , where S ijk For process O ij In processing machine M kThe time to start processing; T ijk The jth process of workpiece i on processing machine M k The time required for processing;

[0015] The workpiece resource constraint means that at any time a workpiece can only be processed by one processing machine or transported by one AGV, which can be expressed as: S ijk +T ijk ≤S i′j′k +L(1-C iji′j′k ),∑A m ,O ij ≤1, where C iji′j′k If process O ij Compared with process O i′j′ First on machine M k If the process is on, it is 1; otherwise, it is 0.

[0016] In response to the flexible job shop scheduling problem with limited processing machine resources and multiple AGVs, the specific technical solution proposed in the present invention is a flexible job shop scheduling method considering limited processing machine resources and multiple AGVs, which includes the following steps:

[0017] S1. Set the problem data and algorithm parameters: including the number of workpieces, the number of processes corresponding to the workpieces, the processing time of the processing machines corresponding to the processes, the total number of processes, the number of processing machines, the number of AGVs, the distance between devices, the average speed of AGVs, the population size Po, the crossover probability Pc, the mutation probability Pm, the number of elites retained m, and the number of iterations I;

[0018] S2, three-segment coding generates the initial population: including,

[0019] S201, using the method of "process code + machine code + AGV code" to represent a set of coding samples, can simultaneously solve the three sub-problems of process processing sequence, process selected processing machine, and process selected AGV. A set of coding samples can be represented as {p1, p2, ..., p n ,m1,m2,…,m n ,a1,a2,…,a n}, p n Indicates the process code, m n Indicates machine code, a n Represents the AGV code, and n is the maximum number of processes.

[0020] S202, initialization parameters, number of iterations s = 1, number of population generation i = 1;

[0021] S203, randomly generate a chromosome X, where the random number selection range of the process code is the number of workpieces, which can be expressed as p n=Range(1,…,i), and the random number of a workpiece in the process code does not exceed the maximum number of processes for the workpiece; the random number selection range of the machine code is the number of processing machines, which can be expressed as m n =Range(1,…,k); The random number selection range of the AGV code is the number of AGVs, which can be expressed as a n =Range(1,…,m);

[0022] S204, determine whether i is equal to the population size Po, if yes, jump to step S205; otherwise i=i+1, jump to step S203;

[0023] S205, generate initial population X init ={X1,X2,…,X Po};

[0024] S3. Evaluate each chromosome using an improved decoding strategy: including,

[0025] S301, determine the workpiece processing sequence by reading the process code, determine the processing machine selected for the process by reading the machine code, and determine the AGV selected for the process by reading the AGV code;

[0026] S302, obtaining the location data of the workpiece warehouse, finished product warehouse, workpieces, processing machines and AGVs, as well as the process processing progress data;

[0027] S303, transporting the workpieces to the loading area of ​​the processing machine in sequence according to the workpiece processing sequence, and determining whether this is the first process. If so, jump to step S304; otherwise, jump to step S305;

[0028] S304. Set the starting point of the transport task to the processing parts warehouse

[0029] S305, setting the starting point of the transport task to the location of the current target workpiece;

[0030] S306, using a greedy algorithm to determine whether the workpiece can be processed in advance during the idle time of the processing machine, if so, jump to step S307; otherwise, jump to step S308;

[0031] S307, setting the transport task destination to the target idle processing machine, and setting the processing task start time to the target idle start time;

[0032] S308, the AGV performs the transport task. At this time, it is necessary to consider the location of the AGV at that time and determine whether the AGV is at the starting point of the transport task. If so, jump to S310; otherwise, jump to step S309;

[0033] S309, dispatching the AGV from its current position to the starting point of the transport task;

[0034] S310, dispatching the AGV to transport the workpiece to the processing machine, and waiting for the processing machine to complete the previous process and start processing the current workpiece before the AGV leaves;

[0035] S311, determine whether the workpiece is in the last process, if so, jump to step S313; otherwise, jump to step S312;

[0036] S312, after the waiting process is completed, jump to step S302;

[0037] S313, occupy the AGV selected for the last process, wait for the process to be completed and then transport the workpiece to the finished product warehouse;

[0038] S4, determine whether the maximum number of iterations of the algorithm I is reached, if yes, jump to step S11; otherwise s=s+1, jump to step S5;

[0039] S5. Calculate the fitness value and sort the population, including:

[0040] S501, calculate the fitness value f of each population according to the objective function of minimizing the maximum completion time, and obtain the fitness value set {f1, ..., f n};

[0041] S502: Sort the population from low to high according to the fitness value to obtain the sorted population set X sort ;

[0042] S6. Select (m+1) chromosomes through the dual strategies of elite retention and tournament selection, including:

[0043] S601, the best chromosome in the sample is directly retained through the elite retention strategy, which can be expressed as

[0044] S602. Select n chromosomes each time using the tournament selection strategy, select the best one and keep it, and keep m chromosomes after m selections.

[0045] S7, alternately select groups and divide the population into groups X according to the fitness value sort By alternating selection, it is divided into a set X1={f1,f3,f5…,f n-1} and set X2={f2,f4,f6…,f n};

[0046] S8, crossover, select parent chromosomes P1 and P2 from set n1 and set X2 respectively according to the crossover probability Pc. The crossover methods include:

[0047] S801: The process code uses workpiece crossover. By randomly selecting a crossover workpiece number set J1 and an unselected workpiece number set J2, the gene sequence containing J1 in the parent generation P1 is copied to the child generation C1, and the gene sequence containing J2 in the parent generation P2 is sequentially inserted into C1. The gene sequence containing J1 in the parent generation P2 is copied to the child generation C2, and the gene sequence containing J2 in the parent generation P1 is sequentially inserted into C2, resulting in two new sets of process codes C1 and C2.

[0048] S802, the machine code and AGV code are sequentially crossed, and the number of crosses N and the cross start position C are selected by random numbers. s The selected sequence cross range must satisfy the constraint condition C s +N≤n, change P1 and P2 from C s Start to C s +N segments of gene sequences are exchanged to obtain offspring C1 and C2;

[0049] S9, mutation, select the parent chromosome P from the chromosome population according to the mutation probability Pm, the mutation methods include:

[0050] S901, the machine code uses random number optimization mutation, selects the mutation position Mu by random number, and then selects the machine with the shortest processing time of the process to obtain the offspring C, which can be expressed as Among them C mu Indicates the mutation position, T ijm Indicates process O ij In processing machine M k The processing time on.

[0051] S902, the process code and AGV code adopt selective mutation, select the mutation positions Mu1 and Mu2 by random numbers, and then exchange the gene sequences of P corresponding to the mutation positions Mu1 and Mu2 to obtain the offspring C;

[0052] S10, update the population, replace the initial population X with the mutated population init The (m+1) chromosomes selected by the elite retention and tournament selection strategies replace the last (m+1) chromosomes in the population sorting to obtain the evolved new population, and then jump to S3;

[0053] S11. Output the optimal factory logistics scheduling plan, calculate the fitness values ​​of all populations, and select the chromosome individual with the smallest fitness value for decoding to obtain the optimal factory logistics scheduling plan.

[0054] The present invention includes an improved genetic algorithm, which generates and retains excellent individuals through the dual strategies of random number optimization mutation and elite retention and tournament selection, thereby improving the algorithm's search efficiency and convergence performance; adding a greedy method in the decoding stage of the algorithm, thereby significantly improving the efficiency of logistics scheduling and reducing production downtime; adopting a three-segment chromosome encoding strategy when encoding the algorithm, so that the solution to each sub-problem can be optimized independently, thereby improving the scalability of the algorithm. This method can solve the problems of poor adaptability of existing flexible job shop scheduling methods in environments with limited processing machine resources and lack of consideration for multi-AGV logistics scheduling, obtain high-quality factory logistics scheduling solutions, and help production decision makers make more reasonable and efficient decisions, thereby reducing production costs and improving production efficiency. Compared with the existing technology, the beneficial effects of the present invention are:

[0055] 1. Aiming at the problems of limited processing machine resources, multiple types of processing machines and complex AGV distribution rules in the production and logistics scheduling of flexible job shops, a mathematical model is established with the constraints of limited processing machine resources, multiple AGV transportation rules and the goal of minimizing the maximum completion time.

[0056] 2. Aiming at the flexible job shop scheduling problem, an improved genetic algorithm is proposed, which adopts a three-segment chromosome encoding strategy to simultaneously solve the three sub-problems of process processing sequence, process selected processing machines, and process selected logistics and transportation equipment. At the same time, the solution of each sub-problem can also be optimized independently, which improves the scalability of the algorithm.

[0057] 3. In response to the constraints of limited processing machine resources and multi-AGV scheduling, a dual strategy of random number optimization mutation and elite retention and tournament selection is adopted to generate and retain excellent individuals, which improves the search efficiency and convergence performance of the algorithm. A decoding method based on the greedy algorithm is designed, which significantly improves the efficiency of logistics scheduling and reduces production downtime. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic diagram of the algorithm flow of the present invention.

[0059] Figure 2 This is an example diagram of the problem of the present invention.

[0060] Figure 3 It is a schematic diagram of the chromosome encoding method of the present invention.

[0061] Figure 4 It is a schematic diagram of the chromosome decoding process of the present invention.

[0062] Figure 5 Schematic diagram of the operation of the greedy decoding algorithm of the present invention.

[0063] Figure 6 It is a scheduling Gantt chart of the problem example of the present invention. DETAILED DESCRIPTION

[0064] The following is a further detailed description of examples of the present invention with reference to the accompanying drawings.

[0065] The present invention is a flexible job shop scheduling method that takes into account the limited resources of processing machines and multiple AGVs. The algorithm flow is as follows: Figure 1 As shown. Figure 2 The following example illustrates the problem.

[0066] S1. Set the problem data and algorithm parameters: number of workpieces: 11, number of processing machines: 24, number of AGVs: 8, total number of processes: 61, average AGV speed: 1.5, population size: 1000, crossover probability: 0.5, mutation probability: 0.1, number of elites retained: 50, number of iterations: 100, and the number of processing machines: the number of workpieces corresponding to the number of processes, and the processing time of the processes corresponding to the processing machines. The distance information between devices is shown in Table 1;

[0067] Table 1. Table of device distance data for the problem example of the present invention

[0068]

[0069]

[0070] S2, three-segment coding generates the initial population: including,

[0071] S201, a set of coding samples is represented by the form of "process code + machine code + AGV code" such as Figure 3 As shown, the three sub-problems of process processing sequence, process selected processing machine, and process selected AGV can be solved simultaneously. A set of coding samples can be represented as {p1, p2, ..., p 61 ,m1,m2,…,m 61 ,a1,a2,…,a 61}, p n Indicates the process code, m n Indicates machine code, a n Indicates the AGV code, and n is the maximum number of processes, which is 61.

[0072] S202, initialization parameters, number of iterations s = 1, number of population generation i = 1;

[0073] S203, randomly generate a chromosome X, where the random number selection range of the process code is the number of workpieces, which can be expressed as p n =Range(1,…,11), and the random number of a workpiece in the process code does not exceed the maximum number of processes for the workpiece; the random number selection range of the machine code is the number of processing machines, which can be expressed as m n=Range(1,…,24); The random number selection range of the AGV code is the number of AGVs, which can be expressed as a n =Range(1,…,8);

[0074] S204, determine whether i is equal to the population size 1000, if yes, jump to step S205; otherwise i=i+1, jump to step S203;

[0075] S205, generate initial population X init ;

[0076] S3. Use the improved decoding strategy to evaluate each chromosome. Figure 4 Shown: including,

[0077] S301, determine the workpiece processing sequence by reading the process code, determine the processing machine selected for the process by reading the machine code, and determine the AGV selected for the process by reading the AGV code;

[0078] S302, obtaining the location data of the workpiece warehouse, finished product warehouse, workpieces, processing machines and AGVs, as well as the process processing progress data;

[0079] S303, transporting the workpieces to the loading area of ​​the processing machine in sequence according to the workpiece processing sequence, and determining whether this is the first process. If so, jump to step S304; otherwise, jump to step S305;

[0080] S304. Set the starting point of the transport task to the processing parts warehouse

[0081] S305, setting the starting point of the transport task to the location of the current target workpiece;

[0082] S306, such as Figure 5 As shown, a greedy algorithm is used to determine whether the workpiece can be processed in advance during the idle time of the processing machine. If so, the process jumps to step S307; otherwise, the process jumps to step S308;

[0083] S307, setting the transport task destination to the target idle processing machine, and setting the processing task start time to the target idle start time;

[0084] S308, the AGV performs the transport task. At this time, it is necessary to consider the location of the AGV at that time and determine whether the AGV is at the starting point of the transport task. If so, jump to step S310; otherwise, jump to step S309;

[0085] S309, dispatching the AGV from its current position to the starting point of the transport task;

[0086] S310, dispatching the AGV to transport the workpiece to the processing machine, and waiting for the processing machine to complete the previous process and start processing the current workpiece before the AGV leaves;

[0087] S311, determine whether the workpiece is in the last process, if so, jump to step S313; otherwise, jump to step S312;

[0088] S312, after the waiting process is completed, jump to step S302;

[0089] S313, occupy the AGV selected for the last process, wait for the process to be completed and then transport the workpiece to the finished product warehouse;

[0090] S4, determine whether the maximum number of algorithm iterations 100 is reached, if yes, jump to step S11; otherwise s=s+1, jump to step S5;

[0091] S5. Calculate the fitness value and sort the population, including:

[0092] S501, calculate the fitness value f of each population according to the objective function of minimizing the maximum completion time, and obtain the fitness value set {f1, ..., f 1000};

[0093] S502: Sort the population from low to high according to the fitness value to obtain the sorted population set X sort ;

[0094] S6, 51 chromosomes were selected through the dual strategies of elite retention and tournament selection, including:

[0095] S601, the best chromosome in the sample is directly retained through the elite retention strategy, which can be expressed as

[0096] S602. Select 20 chromosomes each time using the tournament selection strategy, select the best one and keep it, and keep 50 chromosomes after 50 selections.

[0097] S7, alternate selection grouping, according to the fitness value, the population is divided into a set X1={f1,f3,f5…,f 999} and set X2={f2,f4,f6…,f 1000};

[0098] S8, crossover. Generate a random number between 0 and 1 for each chromosome in sets X1 and X2. Keep the chromosomes in each set with a random number less than 0.5. Then select parent chromosomes P1 and P2 from sets X1 and X2 respectively for crossover to generate daughter chromosomes C1 and C2. This process continues until 500 chromosomes are generated. The crossover methods include:

[0099] S801: The process code uses workpiece crossover. By randomly selecting a crossover workpiece number set J1 and an unselected workpiece number set J2, the gene sequence containing J1 in the parent generation P1 is copied to the child generation C1, and the gene sequence containing J2 in the parent generation P2 is sequentially inserted into C1. The gene sequence containing J1 in the parent generation P2 is copied to the child generation C2, and the gene sequence containing J2 in the parent generation P1 is sequentially inserted into C2, resulting in two new sets of process codes C1 and C2.

[0100] S802, the machine code and AGV code are sequentially crossed, and the number of crosses N and the cross start position C are selected by random numbers. s The selected sequence cross range must satisfy the constraint condition C s +N≤61, change P1 and P2 from C s Start to C s +N segments of gene sequences are exchanged to obtain offspring C1 and C2;

[0101] S9, mutation: Generate a random number between 0 and 1 for each chromosome generated by the crossover. Mutate the chromosomes whose random number is less than 0.1. The mutation methods include:

[0102] S901, the machine code uses random number optimization mutation, selects the mutation position Mu by random number, and then selects the machine with the shortest processing time of the process to obtain the offspring C, which can be expressed as Among them C mu Indicates the mutation position, T ijm Indicates process O ij In processing machine M k The processing time on.

[0103] S902, the process code and AGV code adopt selective mutation, select the mutation positions Mu1 and Mu2 by random numbers, and then exchange the gene sequences of P corresponding to the mutation positions Mu1 and Mu2 to obtain the offspring C;

[0104] S10, update the population, and replace the initial population X with the new chromosome after crossover and mutation init The first 500 chromosomes of the population X are replaced by the 51 chromosomes selected by the elite retention and tournament strategies. init Sort the last 51 chromosomes to obtain the evolved new population and jump to step S3;

[0105] S11. Output the optimal factory logistics scheduling plan, calculate the fitness values ​​of all populations, and select the chromosome individual with the smallest fitness value for decoding to obtain the optimal factory logistics scheduling plan, and draw it into a scheduling Gantt chart as shown below: Figure 6 shown.

[0106] The contents described in the examples of this specification are only used to illustrate the technical solutions of the present invention, rather than to limit them. The scope of protection of the present invention should not be deemed to be limited to the specific forms of implementation described. Those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A flexible job shop scheduling method considering limited processing machine resources and multiple AGVs, characterized by: It includes the following steps: S1. Set the problem data and algorithm parameters: including the number of workpieces, the number of processes corresponding to the workpieces, the processing time of the processing machines corresponding to the processes, the total number of processes, the number of processing machines, the number of AGVs, the distance between devices, the average speed of AGVs, the population size Po, the crossover probability Pc, the mutation probability Pm, the number of elites retained m, and the number of iterations I; S2, three-segment coding to generate the initial population; S3, evaluate each chromosome using the improved decoding strategy; S4, determine whether the maximum number of iterations of the algorithm I is reached, if yes, jump to S11; Otherwise s=s+1, jump to S5; S5, calculate the fitness value and sort the population, include, S51. Calculate the fitness value f of each population according to the objective function of minimizing the maximum completion time, and obtain the fitness value set {f1,…,f n }; S52. Sort the population from low to high according to the fitness value to obtain the sorted population set X sort ; S6. Select (m+1) chromosomes through the dual strategies of elite retention and tournament selection; S7, alternately select groups and divide the population into groups X according to the fitness value sort By alternating selection, it is divided into a set X1={f1,f3,f5…,f n-1 } and set X2={f2,f4,f6…,f n }; S8, crossover, select parent chromosomes P1 and P2 from sets X1 and X2 respectively according to the crossover probability Pc. The crossover methods include: S81, the process code adopts workpiece crossover, by randomly selecting the crossover workpiece number set J1 and the unselected workpiece number set J2, copying the gene sequence containing J1 in the parent generation P1 to the child generation C1, and inserting the gene sequence containing J2 in the parent generation P2 into C1 in sequence, copying the gene sequence containing J1 in the parent generation P2 to the child generation C2, and inserting the gene sequence containing J2 in the parent generation P1 into C2 in sequence, and obtaining two new process codes C1 and C2. 2； S82, machine code and AGV code are sequentially crossed, and the number of crosses N and the cross start position C are selected by random numbers. s The selected sequence cross range must satisfy the constraint condition C s +N≤n, change P1 and P2 from C s Start to C s +N segments of gene sequences are exchanged to obtain offspring C1 and C2; S9, mutation, select the parent chromosome P from the chromosome population according to the mutation probability Pm, and obtain the offspring C through random number optimization mutation and selection mutation; S10, update the population, replace the initial population X with the mutated population init The (m+1) chromosomes selected by the elite retention and tournament selection strategies replace the last (m+1) chromosomes in the population sorting to obtain the evolved new population, and then jump to S3; S11. Output the optimal factory logistics scheduling plan, calculate the fitness values ​​of all populations, and select the chromosome individual with the smallest fitness value for decoding to obtain the optimal factory logistics scheduling plan.

2. A flexible job shop scheduling method considering limited processing machine resources and multiple AGVs according to claim 1, characterized in that: It also includes step S0. Establishing a mathematical model with the goal of limited processing machine resources constraints, multiple AGV transportation rule constraints, and minimizing the maximum completion time, and fully describing the flexible job shop scheduling problem with limited processing machine resources and multiple AGVs.

3. The flexible job shop scheduling method according to claim 1, which takes into account limited processing machine resources and multiple AGVs, is characterized by: The three-stage encoding process of step S2 is as follows: Step S201: A set of code samples is represented by "process code + machine code + AGV code", which can simultaneously solve the three sub-problems of process processing sequence, process selected processing machine, and process selected AGV. A set of code samples can be represented as {p1, p2, ..., p n ,m1,m2,…,m n ,a1,a2,…,a n }, p n Indicates the process code, m n Indicates machine code, a n represents the AGV code, n is the maximum number of processes; Step S202: Initialize parameters, the number of iterations s = 1, the number of populations generated i = 1; Step S203: Randomly generate a chromosome X, where the random number selection range of the process code is the number of workpieces, which can be expressed as p n =Range(1,…,i), and the random number of a workpiece in the process code does not exceed the maximum number of processes for the workpiece; the random number selection range of the machine code is the number of processing machines, which can be expressed as m n =Range(1,…,k); The random number selection range of the AGV code is the number of AGVs, which can be expressed as a n =Range(1,…,m); Sub-step S204: Determine whether i is equal to the population size Po, if yes, skip to sub-step S205; otherwise, i=i+1, skip to sub-step S3; Sub-step S205: Generate initial population X init ={X1,X2,…,X Po }.

4. The flexible job shop scheduling method according to claim 1, which takes into account limited processing machine resources and multiple AGVs, is characterized by: The improved decoding strategy process of step S3 is as follows: Step S301: Determine the workpiece processing sequence by reading the process code, determine the processing machine selected for the process by reading the machine code, and determine the AGV selected for the process by reading the AGV code; Step S302: Obtaining location data of the workpiece warehouse, finished product warehouse, workpieces, processing machines, and AGVs, as well as process processing progress data; Step S303: transport the workpieces to the loading area of ​​the processing machine in sequence according to the workpiece processing sequence, and determine whether this is the first process. If so, jump to step S304; otherwise, jump to step 305; Step S304: Set the starting point of the transport task to the processed parts warehouse Step S305: setting the starting point of the transport task to the location of the current target workpiece; Step S306: Determine whether the workpiece can be processed in advance during the idle time of the processing machine through a greedy algorithm. If so, jump to step 7; otherwise, jump to step S308; Step S307: Set the transport task destination to the target idle processing machine, and set the processing task start time to the target idle start time; Step S308: The AGV performs the transport task. At this time, it is necessary to consider the location of the AGV at that time and determine whether the AGV is at the starting point of the transport task. If so, jump to step S310; otherwise, jump to step S309; Step S309: dispatching the AGV from its current location to the starting point of the transport task; Step S310: The AGV is dispatched to transport the workpiece to the processing machine, and the AGV waits for the processing machine to complete the previous process and start processing the current workpiece before leaving. Step S311: Determine whether the workpiece is in the last process. If so, jump to step S313; otherwise, jump to step S312; Step S312: After the waiting process is completed, jump to step S302; Step S313: Occupy the AGV selected for the last process and wait for the process to be completed before transporting the workpiece to the finished product warehouse.

5. The flexible job shop scheduling method according to claim 1, which takes into account limited processing machine resources and multiple AGVs, is characterized by: The dual strategy process of elite retention and tournament selection in step S6 is as follows: Step S601: directly retain the best chromosome in the sample through the elite retention strategy, which can be expressed as Step S602: n chromosomes are selected each time through the tournament selection strategy, and the best one is selected and retained. After m selections, m chromosomes are retained.

6. The flexible job shop scheduling method according to claim 1, which takes into account limited processing machine resources and multiple AGVs, is characterized by: The random number optimization mutation and selection mutation process in step S9 is as follows: Step S901: The machine code is optimized and mutated using random numbers. The mutation position Mu is selected by random numbers, and then the machine with the shortest processing time of the process is selected to obtain the offspring C, which can be expressed as Among them C mu Indicates the mutation position, T ijm Indicates process O ij In processing machine M k Processing time on Step S902: The process code and AGV code adopt selective mutation, and the mutation positions Mu1 and Mu2 are selected by random numbers. Then, the gene sequences of P corresponding to the mutation positions Mu1 and Mu2 are exchanged to obtain the offspring C.

Images