Multi-AGV flexible job shop active scheduling method based on improved VNS-INSGA-II algorithm and related device

Abstract

The invention provides a multi-AGV flexible job shop active scheduling method based on an improved VNS-INSGA-II algorithm and a related device, and belongs to the technical field of shop active scheduling. According to the method, a multi-objective optimization model for active scheduling of the multi-AGV flexible job shop is constructed, and constraint conditions of the multi-objective optimization model are set; the VNS-INSGA-II algorithm is improved by adopting a multi-layer coding mode, a hybrid population initialization mode, a time-varying coefficient-based dual-strategy binary tournament selection strategy and a variable neighborhood search algorithm, and the improved VNS-INSGA-II algorithm is obtained; based on a set constraint condition of the multi-objective optimization model, an improved VNS-INSGA-II algorithm is adopted to solve decision variables in an objective function of the established multi-objective optimization model for active scheduling of the multi-AGV flexible job shop, and a Pareto frontier is obtained; and selecting a group of solutions from the Pareto frontier to obtain an active scheduling result, thereby carrying out active scheduling on the multi-AGV flexible job shop. According to the invention, the problems of low stability and low scheduling precision of the scheduling system are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of active workshop scheduling, and in particular relates to a multi-AGV flexible workshop active scheduling method and related devices based on an improved VNS-INSGA-II algorithm. Background Art

[0002] In actual production, job-shop scheduling problems are often affected by transportation resources. As an extension of the classic Flexible Job-shop Scheduling Problem (FJSP), FJSP-MA considers both machine and transportation resource constraints, which increases the complexity of the problem.

[0003] The automation upgrade of equipment has exacerbated the fragility of the production system. Machine failure has become the main uncertainty factor in workshop operations, often coupling with the scheduling of multiple AGV (Automated Guided Vehicle) workshops, which will not only disrupt production rhythm but also cause imbalances in material supply.

[0004] Currently, most researchers studying dynamic scheduling use pre-reactive scheduling. This reduces production stability and is susceptible to machine failures, resulting in low system stability and accuracy. Therefore, research into PS-FJSP-MA (Proactive Scheduling in Flexible Job Shops with Multiple AGVs), which takes machine failures into account, meets the needs of actual production. Optimizing resource allocation across the entire shop floor based on predicted machine failure times and skillfully coordinating machine and AGV resources to improve production efficiency, reduce costs and machine load, and mitigate the impact of machine failures on production has become a growing research focus. Summary of the Invention

[0005] The purpose of the present invention is to provide a multi-AGV flexible workshop active scheduling method and related devices based on the improved VNS-INSGA-II algorithm, which are used to solve the problems of low stability and low scheduling accuracy of the scheduling system in the prior art.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a multi-AGV flexible job shop active scheduling method based on an improved VNS-INSGA-II algorithm, comprising the following steps:

[0008] Use machine learning algorithms to predict when machine failures will occur, and define decision variables based on the predicted machine failure times.

[0009] Obtain the working parameters of multi-AGV flexible workshop,

[0010] Based on the working parameters of the multi-AGV flexible job shop and the predicted machine failure time, a multi-objective optimization model for active scheduling of the multi-AGV flexible job shop is constructed. The multi-objective optimization model takes minimizing the maximum completion time, minimizing the total cost, and minimizing the total load as the objective function;

[0011] Setting constraints of the multi-objective optimization model;

[0012] The VNS-INSGA-II algorithm is improved by adopting several layers of encoding, mixed population initialization, dual-strategy binary tournament selection strategy based on time-varying coefficients and variable neighborhood search algorithm. The improved VNS-INSGA-II algorithm is obtained.

[0013] Based on the constraints of the multi-objective optimization model, the improved VNS-INSGA-II algorithm is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible job shops, and the Pareto front is obtained;

[0014] A set of solutions is selected from the Pareto front to obtain active scheduling results, which are used to actively schedule the multi-AGV flexible workshop.

[0015] A further improvement of the present invention is that the objective function of the multi-objective optimization model is expressed as:

[0016]

[0017] Among them, F is the objective function of the multi-objective optimization model, f1 represents minimizing the maximum completion time, f2 represents minimizing the total cost, f3 represents minimizing the total load, and C max is the maximum completion time, PE is the energy consumption cost generated by machine processing, ME is the energy consumption cost generated by the machine when it is idling, RE is the energy consumption cost of AGV loaded, CE is the energy consumption cost of AGV idling, Q max is the total load;

[0018] The calculation formula for the energy consumption cost PE generated by machining is:

[0019]

[0020] The calculation formula for the energy consumption cost ME generated by the machine when it is idling is:

[0021]

[0022] The calculation formula for AGV load energy consumption cost RE is:

[0023]

[0024] The calculation formula for AGV no-load energy consumption cost CE is:

[0025]

[0026] in, For process O ij On machine M k The actual processing time of PC k For machine M k Energy consumption cost per unit time in load state, MC k For machine M k Energy consumption cost per unit time in no-load state, For process O ij On machine M k The processing completion time, For process O ij On machine M k The processing start time, For S h O ij Move to O i,j+1 The end time of the load stroke, For S h O ij Move to O i,j+1 The starting moment of the load stroke, RC h For S h The energy consumption cost per unit time of the load, For S h O i,j+1 Move to O i'j' The end time of the empty load journey, For S h O i,j+1 Move to O i'j' The starting time of the empty load trip, QC h For S h Energy consumption cost per unit time of no-load;

[0027] The decision variables of the objective function of the multi-objective optimization model are:

[0028]

[0029]

[0030] Among them, m k 、 and It is the decision variable designed for active scheduling based on the predicted machine failure time. ijk Assign decision variables to the machine, is the decision variable for process sorting, is the AGV allocation decision variable, is the AGV sorting decision variable;

[0031] The constraints of the multi-objective optimization model include the following formulas:

[0032]

[0033]

[0034] C12:i,i′∈{1,2,L n},j,j′∈{1,2,L u i},M∈{1,2,L m},S∈{1,2,L s}

[0035] Among them, C9 and C10 are machine failure period constraints that need to be avoided during active scheduling.

[0036] A further improvement of the present invention is that the several-layer encoding method specifically includes: using a three-layer encoding method to encode the process layer, the machine layer and the AGV layer respectively.

[0037] A further improvement of the present invention is that the mixed population initialization method specifically includes: randomly generating 10% of the population, locally searching to generate 30% of the population, and globally searching to generate 60% of the population, which together constitute the initial population.

[0038] A further improvement of the present invention is that the calculation formula of the time-varying coefficient is:

[0039] TC=p(x q / (x q +(1-x q )))+1-p

[0040] Where TC is the time-varying coefficient, p and q are the distribution parameter and evolution parameter, respectively.

[0041] A further improvement of the present invention is that the criteria of the tournament selection strategy specifically include:

[0042] Perform a quick non-dominated sort on the initial population and classify the other individuals to obtain multiple non-dominated solution sets.

[0043] Find the highest-rank solution set from multiple non-dominated solution sets;

[0044] The crowding degree of each individual is calculated based on the solution set with the highest rank.

[0045] A further improvement of the present invention is that the variable neighborhood search algorithm specifically includes:

[0046] Perform variable neighborhood search on individuals to obtain variable neighborhood search results;

[0047] The fitness value of the neighborhood solution is calculated based on the variable neighborhood search results.

[0048] In a second aspect, the present invention provides a multi-AGV flexible job shop active scheduling system based on an improved VNS-INSGA-II algorithm, comprising a decision variable definition module, a work parameter acquisition module, a model construction module, a constraint condition setting module, a VNS-INSGA-II algorithm improvement module, a decision variable solution module, and a scheduling module;

[0049] The decision variable definition module is used to use a machine learning algorithm to predict the time when a machine failure occurs, and to define a decision variable based on the predicted time when the machine failure occurs;

[0050] The working parameter acquisition module is used to obtain the working parameters of the multi-AGV flexible operation workshop;

[0051] The model building module is used to build a multi-objective optimization model for active scheduling of a multi-AGV flexible job shop based on the working parameters of the multi-AGV flexible job shop and the predicted machine failure occurrence time. The multi-objective optimization model takes minimizing the maximum completion time, minimizing the total cost and minimizing the total load as objective functions;

[0052] The constraint condition setting module is used to set the constraint conditions of the multi-objective optimization model;

[0053] The VNS-INSGA-II algorithm improvement module is used to improve the VNS-INSGA-II algorithm by adopting several layers of encoding methods, a mixed population initialization method, a dual-strategy binary tournament selection strategy based on time-varying coefficients, and a variable neighborhood search algorithm to obtain an improved VNS-INSGA-II algorithm;

[0054] The decision variable solving module is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible job shops using the improved VNS-INSGA-II algorithm based on the set constraints of the multi-objective optimization model to obtain the Pareto frontier;

[0055] The scheduling module is used to select a set of solutions from the Pareto front to obtain active scheduling results, thereby actively scheduling the multi-AGV flexible workshop.

[0056] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the steps of the above-mentioned method for active scheduling of multi-AGV flexible job shops based on the improved VNS-INSGA-II algorithm.

[0057] In a fourth aspect, the present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned method for active scheduling of multi-AGV flexible job shops based on the improved VNS-INSGA-II algorithm.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The present invention is an improved invention. Compared with the existing active scheduling method for multi-AGV flexible workshops, the present invention improves the VNS-INSGA-II algorithm by using several layers of encoding, a mixed population initialization method, a dual-strategy binary tournament selection strategy based on time-varying coefficients, and a variable neighborhood search algorithm. The improved VNS-INSGA-II algorithm can avoid the defects of the original VNS-INSGA-II algorithm, thereby improving the accuracy of active scheduling of multi-AGV flexible workshops in the later stage. In addition, the loss of machines and AGVs during the production and processing process cannot be ignored. The increase in the load of the machine will cause the machine to malfunction. When establishing the optimization model, the present invention takes the load factor into consideration and uses the total load of the machine (total load) as the scheduling target. When the load of one of the machines is too low, the scheduling system can migrate tasks from the machine that may malfunction to other healthy machines in advance, thereby preventing the scheduling system from being affected by the machine failure and improving the stability of the scheduling system. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 Flowchart of the active scheduling method for multi-AGV flexible job shop based on the improved VNS-INSGA-II algorithm of the present invention;

[0061] Figure 2 Schematic diagram of the multi-AGV flexible job shop active scheduling system based on the improved VNS-INSGA-II algorithm of the present invention;

[0062] Figure 3 This is a flowchart of solving the decision variables in the objective function of the established multi-objective optimization model using the improved VNS-INSGA-II algorithm in Example 3 of the present invention;

[0063] Figure 4 is the Pareto frontier graph of the present invention;

[0064] Figure 5It is the maximum completion time iteration graph of the present invention;

[0065] Figure 6 It is the total cost iteration graph of the present invention;

[0066] Figure 7 It is the total load iteration diagram of the present invention;

[0067] Figure 8 This is a box plot of the completion time solution results of the present invention;

[0068] Figure 9 It is a box plot of the total cost solution results of the present invention;

[0069] Figure 10 It is a box plot of the total load solution results of the present invention;

[0070] Figure 11 The Gantt chart of the scheduling scheme of NSGA-II of the present invention;

[0071] Figure 12 Gantt chart of the scheduling scheme of VNS-INSGA-II of the present invention;

[0072] Figure 13 Actively schedule Gantt chart for the present invention;

[0073] Figure 14 Completely reschedule the Gantt chart for this invention;

[0074] Figure 15 The present invention shifts right and reschedules the Gantt chart;

[0075] Figure 16 Schematic diagram of the structure of the electronic device of the present invention. DETAILED DESCRIPTION

[0076] In order to further understand the content of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the embodiments are only for explaining the present invention and are not intended to limit it.

[0077] The present invention proposes a method for active scheduling of multi-AGV flexible workshops based on the improved VNS-INSGA-II algorithm. This method improves the VNS-INSGA-II algorithm by employing several layers of encoding, a mixed population initialization method, a dual-strategy binary tournament selection strategy based on time-varying coefficients, and a variable neighborhood search algorithm. This method results in an improved VNS-INSGA-II algorithm. Based on the constraints of the multi-objective optimization model, the improved VNS-INSGA-II algorithm is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible workshops, obtaining a Pareto front. A set of solutions is selected from the Pareto front to obtain active scheduling results, thereby performing active scheduling of the multi-AGV flexible workshop. Compared to existing technologies, this method effectively addresses the problems of low scheduling system stability and low scheduling accuracy in existing technologies.

[0078] Example 1:

[0079] The flowchart of the multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm of the present invention is as follows: Figure 1 As shown, the multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm of the present invention includes the following steps:

[0080] S1. Use a machine learning algorithm to predict the time when a machine failure will occur, and define decision variables based on the predicted time when the machine failure will occur.

[0081] S2. Obtain the working parameters of the multi-AGV flexible workshop.

[0082] S3. Based on the working parameters of the multi-AGV flexible job shop and the predicted time of machine failure, a multi-objective optimization model for active scheduling of the multi-AGV flexible job shop is constructed. The multi-objective optimization model takes minimizing the maximum completion time, minimizing the total cost and minimizing the total load as objective functions.

[0083] S4. Setting constraints of the multi-objective optimization model.

[0084] S5. The VNS-INSGA-II algorithm is improved by adopting several layers of encoding, mixed population initialization, dual-strategy binary tournament selection strategy based on time-varying coefficients and variable neighborhood search algorithm to obtain the improved VNS-INSGA-II algorithm.

[0085] S6. Based on the constraints of the multi-objective optimization model, the improved VNS-INSGA-II algorithm is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible job shops to obtain the Pareto frontier.

[0086] S7. Select a set of solutions from the Pareto front to obtain active scheduling results, and use them to actively schedule the multi-AGV flexible workshop.

[0087] Example 2:

[0088] The schematic diagram of the multi-AGV flexible workshop active scheduling system based on the improved VNS-INSGA-II algorithm of the present invention is as follows: Figure 2 As shown, the multi-AGV flexible job shop active scheduling system based on the improved VNS-INSGA-II algorithm of the present invention includes a decision variable definition module, a working parameter acquisition module, a model construction module, a constraint condition setting module, a VNS-INSGA-II algorithm improvement module, a decision variable solving module and a scheduling module.

[0089] The decision variable definition module is used to use a machine learning algorithm to predict the time when a machine failure occurs, and to define decision variables based on the predicted time when a machine failure occurs.

[0090] The working parameter acquisition module is used to obtain the working parameters of the multi-AGV flexible operation workshop.

[0091] The model building module is used to construct a multi-objective optimization model for active scheduling of multi-AGV flexible job shops based on the working parameters of the multi-AGV flexible job shops and the predicted time of machine failure. The multi-objective optimization model takes minimizing the maximum completion time, minimizing the total cost and minimizing the total load as the objective functions.

[0092] The constraint condition setting module is used to set the constraint conditions of the multi-objective optimization model.

[0093] The VNS-INSGA-II algorithm improvement module is used to improve the VNS-INSGA-II algorithm by adopting several layers of encoding methods, mixed population initialization methods, dual-strategy binary tournament selection strategy based on time-varying coefficients and variable neighborhood search algorithm to obtain the improved VNS-INSGA-II algorithm.

[0094] The decision variable solving module is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible job shops based on the set constraints of the multi-objective optimization model, using the improved VNS-INSGA-II algorithm to obtain the Pareto frontier.

[0095] The scheduling module is used to select a set of solutions from the Pareto front to obtain active scheduling results, thereby actively scheduling the multi-AGV flexible workshop.

[0096] Example 3:

[0097] The present invention provides a multi-AGV flexible job shop active scheduling method based on an improved VNS-INSGA-II algorithm, comprising the following steps:

[0098] S1. Use a machine learning algorithm to predict the time when a machine failure will occur, and define decision variables based on the predicted time when the machine failure will occur.

[0099] A machine learning algorithm is used to predict the time when a machine failure will occur. Decision variables are defined based on the predicted time when the machine failure will occur. The defined decision variables are as follows:

[0100]

[0101]

[0102] Among them, m k 、 and It is the decision variable designed for active scheduling based on the predicted machine failure time. ijk Assign decision variables to the machine, is the decision variable for process sorting, is the AGV allocation decision variable, is the AGV sorting decision variable.

[0103] S2. Obtain the working parameters of the AGV flexible operation workshop.

[0104] First, the working parameters of the AGV flexible operation workshop are obtained. The obtained working parameters of the AGV flexible operation workshop are shown in Table 1.

[0105] Table 1 Parameter meaning

[0106]

[0107]

[0108] S3. Based on the working parameters of the multi-AGV flexible job shop and the predicted machine failure time, a multi-objective optimization model for active scheduling of the multi-AGV flexible job shop is constructed.

[0109] Firstly, based on the working parameters of the multi-AGV flexible job workshop and the predicted machine failure time, a multi-objective optimization model for active scheduling of the multi-AGV flexible job workshop is constructed. The multi-objective optimization model for active scheduling of the multi-AGV flexible job workshop takes minimizing the maximum completion time, minimizing the total cost and minimizing the total load as the objective function.

[0110] The objective function of the multi-objective optimization model is introduced in detail below. The expression of the objective function of the multi-objective optimization model is:

[0111]

[0112] Among them, F is the objective function of the multi-objective optimization model, f1 represents minimizing the maximum completion time, f2 represents minimizing the total cost, f3 represents minimizing the total load, and C max is the maximum completion time, PE is the energy consumption cost generated by machine processing, ME is the energy consumption cost generated by the machine when it is idling, RE is the energy consumption cost of AGV loaded, CE is the energy consumption cost of AGV idling, Q max is the total load.

[0113] The calculation formula for the energy consumption cost PE generated by machining is:

[0114]

[0115] The calculation formula for the energy consumption cost ME generated by the machine when it is idling is:

[0116]

[0117] The calculation formula for AGV load energy consumption cost RE is:

[0118]

[0119] The calculation formula for AGV no-load energy consumption cost CE is:

[0120]

[0121] S4. Set constraints for the multi-objective optimization model.

[0122] The constraints of the multi-objective optimization model include the following formulas:

[0123]

[0124] C12:i,i′∈{1,2,L n},j,j′∈{1,2,L u i},M∈{1,2,L m},S∈{1,2,L s}

[0125] Among them, C9 and C10 are machine failure period constraints that need to be avoided during active scheduling.

[0126] The constraints are described in detail below:

[0127] Specifically, C1 indicates that processing operations cannot be interrupted, ensuring the priority relationship between operations. C2 indicates that each process can only be assigned to a single machine during processing. C3 indicates that the transfer task of each process can only be performed by one AGV. C4 indicates that conflicts in the order of processes assigned to machines should be avoided. C5 indicates that conflicts in the AGV transport sequence should be avoided. C6 indicates that the start time of any process must be after the end of the load travel and the completion time of the previous process on the same machine. C7 indicates that the start time of the AGV empty transport must be after the completion of the load task of the previous process and the start time of the process of the workpiece to be transported. C8 indicates that the start time of the AGV loaded transport must be after the completion time of the workpiece and the end time of the AGV empty transport. C9 indicates that the process must be processed either before or after the machine failure. C10 indicates that the process planned to be processed on the faulty machine must be completed no later than the time the fault occurs, or its start time must be no earlier than the time the fault is repaired. C11 indicates the value range of the decision variable. C12 indicates the value range of the parameter.

[0128] S5. The VNS-INSGA-II algorithm is improved by adopting several layers of encoding, mixed population initialization, dual-strategy tournament selection strategy based on time-varying coefficients and variable neighborhood search algorithm to obtain the improved VNS-INSGA-II algorithm.

[0129] The several-layer coding method specifically includes: using a three-layer coding method to encode the process layer, machine layer and AGV layer respectively.

[0130] The mixed population initialization method specifically includes: randomly generating 10% of the population, locally searching to generate 30% of the population, and globally searching to generate 60% of the population, which together constitute the initial population.

[0131] The calculation formula of the time-varying coefficient is:

[0132] TC=p(x q / (x q +(1-x q )))+1-p

[0133] Where TC is the time-varying coefficient, p and q are the distribution parameter and evolution parameter, respectively.

[0134] The specific guidelines for tournament selection strategy include:

[0135] A. Perform a quick non-dominated sort on the initial population and classify the other individuals to obtain multiple non-dominated solution sets.

[0136] B. Find the highest-rank solution set from multiple non-dominated solution sets;

[0137] C. Calculate the crowding degree of each individual based on the highest-rank solution set.

[0138] The variable neighborhood search algorithm specifically includes:

[0139] a. Perform variable neighborhood search on the individual to obtain variable neighborhood search results;

[0140] b. Calculate the fitness value of the neighborhood solution based on the variable neighborhood search results.

[0141] S6. Based on the constraints of the set multi-objective optimization model, the improved VNS-INSGA-II algorithm is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible job shops, and the Pareto front is obtained.

[0142] In this step, the improved VNS-INSGA-II algorithm is used to solve the decision variables in the objective function of the established multi-objective optimization model. Figure 3 The specific process is as follows:

[0143] Step 1: Chromosome encoding and decoding: Using a three-layer encoding scheme of machine allocation (MS), process sorting (OS), and AGV selection (AS), a "first-come, first-served" strategy is designed to decode the dyeing operation;

[0144] Step 2: Use a hybrid search population initialization method based on load balancing: 10% of the initial population is random search, 60% of the initial population is global search, and 20% of the initial population is local search to generate the initial population P0. max , crossover probability P c and mutation probability P m Perform parameter initialization settings;

[0145] Step 3: Fast non-dominated sorting and calculation of crowding: Calculate the fitness and fast non-dominated sorting of the initial population, classify the remaining individuals, obtain multiple non-dominated solution sets, find the solution set with the highest level, calculate the crowding degree of each individual, and generate the next generation of elite individuals;

[0146] Step 4: Use the dual-strategy tournament strategy based on time-varying coefficients to select the best parent individuals in the population;

[0147] Step 5: Crossover and mutation operations: Generate offspring individuals through crossover operations, and perform mutation operations on the generated offspring individuals to generate new offspring individuals;

[0148] Step 6: Perform variable neighborhood search on the individual, calculate the fitness value of the neighborhood solution, and cover the original chromosome with the better solution;

[0149] Step 7: Merge populations: select a new parent population through the elite retention strategy, which is the population after iteration;

[0150] Step 8: Determine whether the number of iterations is met: If so, output the optimal solution; otherwise, repeat Step 4 to Step 7.

[0151] Specifically, the encoding is as follows:

[0152] The OS layer, the machine MS layer, and the AGV AS layer are encoded separately, each with a length of 3u, where u represents the number of processes. The OS layer is randomly generated using the serial number of each workpiece. The gene value represents the workpiece number, and the order of appearance indicates the workpiece's process. The MS and AS layers are encoded using the order of the workpieces, encoding the processing machine and AGV selected for each process from left to right.

[0153] The specific process of the "preconceived" decoding strategy in step S1 is as follows:

[0154] a. Decoding chromosome information. Extract the genes in the corresponding chromosome from the process sequence and convert them into the corresponding process. ij , extract the machine M corresponding to the process from the machine allocation chromosome gene k , extract the corresponding AGVS from the AGV gene sequence h , record the machine failure time window constraints.

[0155] b. Update process information. Find the processing time of the process on the machine and update the process information.

[0156] c. Process the first process and AGV transport. Determine whether the current process is the first processing process. If so, and the AGV is currently at the loading station, its empty transport start time is 0. If the AGV is not at the loading station, its empty transport time is the transport time from its current location to the target machine, and its loaded transport time is the transport time from the target machine to the next processing process. If the current process is not the first process, proceed to Step 4.

[0157] d. Handling AGV transport between two consecutive processes. Determine whether the two consecutive processes of the workpiece are processed on the same machine. If so, the AGV's empty and loaded transport times are both zero. Otherwise, the AGV must move from its current location to the designated location, waiting for the process to complete before being loaded for transport. In this case, the AGV's empty time is the time required to move to the designated location.

[0158] e. Determine the transport AGV. Based on the transport schedule between machines, calculate the arrival time of all AGVs at the processing machine for the current process. If multiple AGVs arrive later than the process completion time, the earliest arriving AGV is selected for transport; otherwise, the latest arriving AGV is selected for transport.

[0159] f. Determine whether there is a faulty machine. If so, avoid the faulty machine's maintenance time based on the constraints. Otherwise, ignore this step.

[0160] g. Calculate transport time and process time. Based on the transport schedule between machines, calculate the time it takes for all AGVs to arrive at the machine for the current process. If multiple AGVs arrive later than the process's completion time, the earliest arriving AGV is selected for transport; otherwise, the latest arriving AGV is selected for transport.

[0161] h. Generate a scheduling plan. Based on the above steps, determine the processing order for each workpiece and select the appropriate AGV to transport the workpiece. Finally, generate a feasible scheduling plan that includes processing and transportation information.

[0162] S7. Select a set of solutions from the Pareto front to proactively schedule the multi-AGV flexible job shop.

[0163] To verify the performance of the proposed method (the improved VNS-INSGA-II algorithm, referred to as VNS-INSGA-II), this example constructs RMK01-RMK10 cases based on the MK01-MK10 cases. The transfer times between different machines are randomly generated between [1, 2]. The corresponding machine processing energy consumption and AGV transport energy consumption are designed. The predicted faulty machine and fault onset time are randomly set for each case. The scale of the RMK01-RMK10 cases is shown in Table 2.

[0164] Table 2 Calculation scale of RMK01 to RMK10

[0165]

[0166] This example uses this example to solve the problem. Under the same environment, the NSGA-II algorithm and the VNS-INSGA-II algorithm are run 10 times, and the average and optimal values ​​are obtained. Table 3 shows the objective function values, algorithm execution time, and optimization rate of the two algorithms. The values ​​with good performance are bolded. It can be seen that the VNS-INSGA-II algorithm has better solution results and longer algorithm runtime.

[0167] Table 3 Objective function values ​​and optimization rates solved by two algorithms

[0168]

[0169]

[0170] This embodiment takes RMK01 as an example. Figure 4 The Pareto frontier distribution plots for the two algorithms are shown in Figure 2. From the projection, we can observe a significant negative correlation between total load and completion time, indicating that a decrease in completion time leads to an increase in total machine load. Total cost and completion time exhibit a positive correlation, with an increase in completion time leading to an increase in total cost. The non-dominated solutions obtained by the VNS-INSGA-II algorithm are more concentrated in the lower left region of the frontier distribution, further verifying the superiority of the VNS-INSGA-II algorithm.

[0171] Select the individual with the shortest completion time in the Pareto frontier level and compare and analyze the iterative convergence process of the algorithm when solving the three objective functions. Figures 5 to 7 As shown in Figure 2, the VNS-INSGA-II algorithm converges faster and has better solution quality. To further verify the stability of the algorithm, a box plot of the objective function solution results is drawn based on 10 independent experiments, as shown in Figure 2. Figures 6 to 8 As shown in the figure, both algorithms have good stability. Compared with the NSGA-II algorithm, the average and median of the VNS-INSGA-II algorithm are lower and the stability is better.

[0172] Table 4 shows the average results of the HV (Hypervolume) and IGD (Inverted Generational Distance) metrics for the two algorithms on 10 benchmark cases. The higher performing values ​​are highlighted in bold. The VNS-INSGA-II algorithm achieved 10 HV-optimal examples, while the NSGA-II algorithm achieved only 0. The VNS-INSGA-II algorithm achieved 9 IGD-optimal examples, while the NSGA-II algorithm achieved only 1. In summary, the VNS-INSGA-II algorithm performed better in both HV and IGD metrics.

[0173] Table 4 Algorithm index evaluation results

[0174]

[0175] As shown in Table 4, the VNS-INSGA-II algorithm can obtain 10 HV-excellent examples, while the NSGA-II algorithm has only 0 HV-excellent examples. The VNS-INSGA-II algorithm can obtain 9 IGD-excellent examples, while the NSGA-II algorithm has only 1 IGD-excellent example. In summary, the VNS-INSGA-II algorithm performs better in terms of HV and IGD indicators.

[0176] This example takes RMK01 as an example and selects the solution with the shortest completion time. Figure 11 and 12 As shown in the figure, the completion time of the VNS-INSGA-II algorithm is 62.3 minutes, which is better than the 65.4 minutes of the NSGA-II algorithm.

[0177] The proposed active scheduling model was compared with two passive scheduling methods commonly used to address machine failure interference (full rescheduling and right-shift rescheduling) to verify the superiority of the active scheduling model. The solution was obtained using the highly performing VNS-INSGA-II algorithm. The example was executed 10 times and the average and optimal values ​​were taken. The target values ​​and execution times for active scheduling, full rescheduling, and right-shift rescheduling are shown in Table 5, with the superior values ​​highlighted in bold.

[0178] Table 5 Target values ​​and running times of different dynamic scheduling methods

[0179]

[0180]

[0181] Table 5 shows that, among the solution results for cases RMK01 to RMK10, active scheduling achieved optimal performance in nine cases, while full rescheduling achieved optimal performance in only one case, and the right-shift rescheduling strategy achieved no optimal performance. Active scheduling achieved superior results compared to full rescheduling and right-shift rescheduling. Right-shift rescheduling yielded inferior results but the shortest runtime.

[0182] This embodiment takes RMK01 as an example, the predicted faulty machine is M2, the fault start time is 35 minutes, and the fault repair time is 10 minutes. Figures 13 to 15 As can be seen from the figure, the completion time of the active scheduling is 66 minutes, which is better than the 75 minutes of the complete rescheduling solution and the 87 minutes of the right-shift rescheduling solution.

[0183] Example 4:

[0184] See also Figure 16 As shown, the present invention also provides an electronic device 100 for a multi-AGV flexible job shop active scheduling method based on an improved VNS-INSGA-II algorithm; the electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104.

[0185] The memory 101 can be used to store the computer program 103. The processor 102 implements the steps of the multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm described in Example 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data (such as audio data) created according to the use of the electronic device 100. In addition, the memory 101 can include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device.

[0186] The at least one processor 102 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 may be a microprocessor or any conventional processor, etc. The processor 102 is the control center of the electronic device 100 and connects various parts of the entire electronic device 100 using various interfaces and lines.

[0187] The memory 101 in the electronic device 100 stores a plurality of instructions to implement a multi-AGV flexible job shop active scheduling method based on an improved VNS-INSGA-II algorithm. The processor 102 can execute the plurality of instructions to implement:

[0188] Use machine learning algorithms to predict when machine failures will occur, and define decision variables based on the predicted machine failure times.

[0189] Obtain working parameters of multi-AGV flexible workshop;

[0190] Based on the working parameters of the multi-AGV flexible job shop and the predicted machine failure time, a multi-objective optimization model for active scheduling of the multi-AGV flexible job shop is constructed. The multi-objective optimization model takes minimizing the maximum completion time, minimizing the total cost, and minimizing the total load as the objective function;

[0191] Setting constraints of the multi-objective optimization model;

[0192] The improved VNS-INSGA-II algorithm is obtained by adopting a dual-strategy tournament selection strategy based on time-varying coefficients.

[0193] Based on the constraints of the multi-objective optimization model, the improved VNS-INSGA-II algorithm is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible job shops, and the Pareto front is obtained;

[0194] A set of solutions is selected from the Pareto front to obtain active scheduling results, which are used to actively schedule the multi-AGV flexible workshop.

[0195] Example 5:

[0196] If the module / unit integrated in the electronic device 100 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory and read-only memory (ROM, Read-Only Memory).

[0197] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0198] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0199] These computer program instructions can also be stored in a computer readable memory that can guide a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 One or more boxes specify a function.

[0200] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0201] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm, characterized by: The following steps are involved: Use machine learning algorithms to predict when machine failures will occur, and define decision variables based on the predicted machine failure times. Obtain working parameters of multi-AGV flexible workshop; Based on the working parameters of the multi-AGV flexible job shop and the predicted machine failure time, a multi-objective optimization model for active scheduling of the multi-AGV flexible job shop is constructed. The multi-objective optimization model takes minimizing the maximum completion time, minimizing the total cost, and minimizing the total load as the objective function; Setting constraints of the multi-objective optimization model; The VNS-INSGA-II algorithm is improved by adopting several layers of encoding, mixed population initialization, dual-strategy binary tournament selection strategy based on time-varying coefficients and variable neighborhood search algorithm. The improved VNS-INSGA-II algorithm is obtained. Based on the constraints of the multi-objective optimization model, the improved VNS-INSGA-II algorithm is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible job shops, and the Pareto front is obtained; A set of solutions is selected from the Pareto front to obtain active scheduling results, which are used to actively schedule the multi-AGV flexible workshop.

2. The multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm according to claim 1 is characterized in that: The expression of the objective function of the multi-objective optimization model is: Among them, F is the objective function of the multi-objective optimization model, f1 represents minimizing the maximum completion time, f2 represents minimizing the total cost, f3 represents minimizing the total load, and C max is the maximum completion time, PE is the energy consumption cost generated by machine processing, ME is the energy consumption cost generated by the machine when it is idling, RE is the energy consumption cost of AGV loaded, CE is the energy consumption cost of AGV idling, Q max is the total load; The calculation formula for the energy consumption cost PE generated by machining is: The calculation formula for the energy consumption cost ME generated by the machine when it is idling is: The calculation formula for AGV load energy consumption cost RE is: The calculation formula for AGV no-load energy consumption cost CE is: in, For process O ij On machine M k The actual processing time of PC k For machine M k Energy consumption cost per unit time in load state, MC k For machine M k Energy consumption cost per unit time in no-load state, For process O ij On machine M k The processing completion time, For process O ij On machine M k The processing start time, For S h O ij Move to O i,j+1 The end time of the load stroke, For S h O ij Move to O i,j+1 The starting moment of the load stroke, RC h For S h The energy consumption cost per unit time of the load, For S h O i,j+1 Move to O i'j' The end time of the empty load journey, For S h O i,j+1 Move to O i'j' The starting time of the empty load trip, QC h For S h Energy consumption cost per unit time of no-load; The decision variables of the objective function of the multi-objective optimization model are: Among them, m k 、 and It is the decision variable designed for active scheduling based on the predicted machine failure time. ijk Assign decision variables to the machine, is the decision variable for process sorting, is the AGV allocation decision variable, is the AGV sorting decision variable; The constraints of the multi-objective optimization model include the following formulas: C12:i,i′∈{1,2,L n},j,j′∈{1,2,L u i },M∈{1,2,L m},S∈{1,2,L s} Among them, C9 and C10 are machine failure period constraints that need to be avoided during active scheduling.

3. The multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm according to claim 1 is characterized in that: The several-layer encoding method specifically includes: using a three-layer encoding method to encode the process layer, the machine layer and the AGV layer respectively.

4. The multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm according to claim 1 is characterized in that: The mixed population initialization method specifically includes: randomly generating 10% of the population, locally searching to generate 30% of the population, and globally searching to generate 60% of the population, which together constitute the initial population.

5. The multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm according to claim 1 is characterized in that: The calculation formula of the time-varying coefficient is: TC=p(x q / (x q +(1-x q )))+1-p Where TC is the time-varying coefficient, p and q are the distribution parameter and evolution parameter, respectively.

6. The multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm according to claim 1 is characterized in that: The criteria for the tournament selection strategy specifically include: Perform a quick non-dominated sort on the initial population and classify the other individuals to obtain multiple non-dominated solution sets. Find the highest-rank solution set from multiple non-dominated solution sets; The crowding degree of each individual is calculated based on the solution set with the highest rank.

7. The multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm according to claim 1 is characterized in that: The variable neighborhood search algorithm specifically includes: Perform variable neighborhood search on individuals to obtain variable neighborhood search results; The fitness value of the neighborhood solution is calculated based on the variable neighborhood search results.

8. A multi-AGV flexible job shop active scheduling system based on the improved VNS-INSGA-II algorithm, characterized by: It includes decision variable definition module, working parameter acquisition module, model construction module, constraint condition setting module, VNS-INSGA-II algorithm improvement module, decision variable solution module and scheduling module; The decision variable definition module is used to use a machine learning algorithm to predict the time when a machine failure occurs, and to define a decision variable based on the predicted time when the machine failure occurs; The working parameter acquisition module is used to obtain the working parameters of the multi-AGV flexible operation workshop; The model building module is used to build a multi-objective optimization model for active scheduling of a multi-AGV flexible job shop based on the working parameters of the multi-AGV flexible job shop and the predicted machine failure occurrence time. The multi-objective optimization model takes minimizing the maximum completion time, minimizing the total cost and minimizing the total load as objective functions; The constraint condition setting module is used to set the constraint conditions of the multi-objective optimization model; The VNS-INSGA-II algorithm improvement module is used to improve the VNS-INSGA-II algorithm by adopting several layers of encoding methods, a mixed population initialization method, a dual-strategy binary tournament selection strategy based on time-varying coefficients, and a variable neighborhood search algorithm to obtain an improved VNS-INSGA-II algorithm; The decision variable solving module is used to solve the decision variables in the objective function of the established multi-objective optimization model for active scheduling of multi-AGV flexible job shops using the improved VNS-INSGA-II algorithm based on the set constraints of the multi-objective optimization model to obtain the Pareto frontier; The scheduling module is used to select a set of solutions from the Pareto front to obtain active scheduling results, thereby actively scheduling the multi-AGV flexible workshop.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm are implemented as described in any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-AGV flexible job shop active scheduling method based on the improved VNS-INSGA-II algorithm according to any one of claims 1 to 7 are implemented.

Images