A flexible job shop energy-saving batch flow scheduling optimization method considering machine deterioration and preventive maintenance

Abstract

The present application relates to a kind of flexible job shop energy-saving batch flow scheduling optimization method considering machine tool deterioration and preventive maintenance, belong to workshop scheduling technical field, including: the construction of considering machine tool deterioration and preventive maintenance of flexible job shop energy-saving batch flow scheduling mathematical model with completion time, total energy consumption and total production cost as optimization goal;Propose model solution method, introduce strategy based on the population diversity of historical population, enhance the response capability of algorithm to environmental change;Develop adaptive crossover mutation operator, balance the diversity and convergence of algorithm;Propose neighborhood search strategy, explore the local optimal solution of energy-saving batch flow scheduling problem.The present application can efficiently solve the multi-objective optimization problem of considering machine tool deterioration and preventive maintenance of flexible job shop energy-saving batch flow scheduling, effectively avoid machine tool under-maintenance or over-maintenance, reduce production interruption and maintenance cost, obtain the scheduling scheme with good completion time, total energy consumption and total production cost.

Description

TECHNICAL FIELD

[0001] The application relates to a flexible job shop energy-saving batch flow scheduling optimization method considering machine tool deterioration and preventive maintenance, and belongs to the technical field of shop scheduling. BACKGROUND

[0002] Machine tools are key basic equipment in a workshop, and their states and performances directly affect the execution of production scheduling schemes. In actual production in a workshop, key components of machine tools are affected by factors such as temperature effect, vibration impact, load change and working environment when running for a long time, and thus wear and fatigue occur, the precision of the machine tools decreases and the performance deteriorates, and thus the machining time of the machine tools changes. The influence of the deterioration effect on workpiece production scheduling will continue to intensify as the damage accumulates, which not only leads to an increase in completion time, energy consumption and cost, but also interferes with the normal production scheduling plan, and therefore, maintenance must be performed in time. Although traditional periodic maintenance is simple and easy to implement, it is not necessary every time, which leads to a waste of a large amount of resources such as cost and time. Compared with the traditional periodic maintenance, non-periodic preventive maintenance is more flexible and practical, can effectively avoid under-maintenance or over-maintenance of machine tools, and reduce production interruption and maintenance cost. Therefore, it is particularly important to study an optimization method for energy-saving batch flow scheduling in a flexible job shop considering machine tool deterioration and preventive maintenance.

[0003] The deterioration effect of machine tools leads to uncertainty of workpiece machining time, and thus the Pareto optimal solution set in the decision space and the Pareto front in the target space of the energy-saving batch flow scheduling problem in a flexible job shop change with the scheduling, and thus the robustness of traditional optimization methods is insufficient. Meanwhile, when the deterioration accumulates to a certain degree, preventive maintenance is needed, and under the flexible maintenance environment, the convergence degree of the target values of the optimized scheduling schemes obtained based on different maintenance intervals is very different due to the constraints of various production resources in the workshop. Therefore, how to reasonably represent the deterioration effect of machine tools, and on the basis, how to construct an energy-saving batch flow scheduling mathematical model of a flexible job shop considering machine tool deterioration and preventive maintenance are key to realizing robust optimization of the energy-saving batch flow scheduling of a flexible job shop considering machine tool deterioration and preventive maintenance.

[0004] At present, scholars have carried out relevant researches on single machine / parallel machine deteriorating scheduling, mixed flow shop / flexible job shop deteriorating scheduling and deteriorating scheduling considering workpiece batching from aspects of model construction and optimization algorithm design, and have achieved certain results. However, at present, no relevant researches considering machine tool deterioration, preventive maintenance and flexible production batch flow scheduling are searched; for the flexible job shop energy-saving batch flow scheduling problem, the workpiece batching will lead to the influence of workpiece processing time uncertainty on the workshop scheduling more prominent with the increase of problem scale, and the existing researches on machine tool deterioration effect are mostly based on linear (step) function under ideal conditions, which cannot accurately depict the influence of machine tool deterioration on processing time in the real production environment, so that the obtained optimized scheduling scheme lacks practical guiding significance. Meanwhile, under the competition of various manufacturing resources in the workshop, the selection of preventive maintenance interval also has an important influence on the performance of the scheduling scheme. Therefore, how to reasonably represent the influence of machine tool deterioration on workpiece processing time, and on this basis, establish a flexible job shop energy-saving batch flow scheduling mathematical model considering machine tool deterioration and preventive maintenance and design an efficient and feasible optimization algorithm still need further exploration. SUMMARY

[0005] In view of the above deficiencies and improvement requirements of the prior art, the present application provides a flexible job shop energy-saving batch flow scheduling optimization method considering machine tool deterioration and preventive maintenance, which aims to solve the problem of insufficient robustness of traditional optimization methods caused by machine tool deterioration and flexible maintenance, and realize the flexible job shop energy-saving batch flow multi-objective robust optimization scheduling considering machine tool deterioration and preventive maintenance.

[0006] To solve the above technical problems, the present application provides a flexible job shop energy-saving batch flow scheduling optimization method considering machine tool deterioration and preventive maintenance, comprising the following steps: S1, constructing a flexible job shop energy-saving batch flow scheduling mathematical model considering machine tool deterioration and preventive maintenance with the optimization objectives of completion time, total energy consumption and total production cost; S2, proposing a hybrid method of knowledge-driven batching method and dynamic self-adaptive multi-objective evolutionary algorithm to solve the model; S3, designing a problem knowledge-driven neighborhood strategy to further explore the local optimal solution of the problem.

[0007] According to the present application, as a further preferred, the flexible job shop energy-saving batch flow scheduling multi-objective optimization mathematical model considering machine tool deterioration and preventive maintenance in the step S1, the optimization objectives of the model are the shortest completion time, the lowest total energy consumption and the lowest total production cost, and the calculation formula is:

[0008] F = min (T e , T m , T c )

[0009] Wherein, F is the objective function, Te T m T c These are the total energy consumption of the workshop, the completion time, and the total production cost.

[0010] T e =E c +E ct +E id +E m +E f

[0011]

[0012]

[0013]

[0014]

[0015] E f =E u ×T max

[0016] Among them, E c E ct E id E m E f and E u These represent processing energy consumption, processing preparation energy consumption, idle energy consumption, transportation energy consumption, stationary energy consumption, and common energy consumption generated per unit time in the workshop; i, j, k, and r are the indices of the workpiece, process, machine tool, and tooling, respectively; n', N t q i m represents the total number of workpieces after batching, and J represents the number of workpieces. i Batch size, tooling quantity, and workpiece J of the lth sub-batch i The number of processes and the number of machine tools; and P m Machine tool M k The standby power, idle power, and AGV handling power; and They are respectively process Q i,j In machine tool M k The time required for clamping, changing tools, adjusting tools, and transferring processes (Q) i,j' The time required for transporting the equipment to the new machine tool for processing; X i,j,k Y i,j,k,r Z i,j and MI k As a decision variable, when process Q i,j In machine tool M k X during processingi,j,k 1, otherwise 0, if the process Q i,j In the machine tool M k The processing uses tooling MT r Y i,j,k,r 1, otherwise 0, if the process Q i,j and its next process are processed on the same machine tool Z i,j 1, otherwise 0, if the machine tool M k is available at the scheduling time MI k 1, otherwise 0.

[0017]

[0018] Where, is the end time of the q i th process of the workpiece J i on the machine tool M k .

[0019] The total production cost of the workpiece is composed of the processing cost of the workpiece and the maintenance cost of the machine tool, which is represented as:

[0020]

[0021] Where, C i,j,k , MC k and NC k are the unit time processing cost of the process Q i,j on the machine tool M k , the maintenance frequency of the machine tool M k and the average cost of the machine tool M k maintenance once.

[0022] In addition to meeting the basic constraint conditions of multi-resource flexible job shop batch flow scheduling, the machine tool deterioration constraint needs to be met, which is represented as:

[0023]

[0024] Where, D k is the deterioration coefficient, IT k is the processing time of the machine tool M k before triggering maintenance, is the initial processing time of Q i,j on the machine tool M k .

[0025] Machine tool available time window constraint. When triggering preventive maintenance, if there is a processing task on the machine tool that has not been completed, it needs to wait for its completion before maintenance, which is represented as:

[0026]

[0027]

[0028] wherein, is a decision variable, taking value 1 if the machine M k is the next operation Q ij after the operation Q gh , and 0 otherwise; A i,j and A g,h are the sets of machines M ij and M gh that can process the operations Q k and Q k , respectively; W k is a decision variable, taking value 1 if the machine M k triggers maintenance, and 0 otherwise; is the starting time point of the b-th maintenance of the machine M k ; LBD k and UPD k are the minimum and maximum deterioration of the machine M k , respectively, when it is under maintenance.

[0029] Machine maintenance constraints. A preventive maintenance can be triggered multiple times within a certain production period, and the ending time point of the b-th maintenance of the machine M k is denoted as:

[0030]

[0031] wherein, MT k is the time required for the maintenance of the machine M k .

[0032] According to the present application, as a further preferred, the step S2 comprises the following sub-steps:

[0033] Design a population diversity preservation strategy based on historical population. By fusing the historical population into the current population in a certain proportion in each search stage, the diversity of the current population is improved. The specific operation is as follows:

[0034] Step 1: use MFO operator to optimize the initialized population Population to obtain the offspring population Population2;

[0035] Step 2: fuse Population and Population2 in a 1:1 proportion, and select the merged population based on IGD-NS selection strategy to obtain Population3;

[0036] Step 3: optimize the population Population3 based on the synergistic evolution operator to obtain the offspring population Population4;

[0037] Step 4: Randomly initialize the population to obtain Population 5, fuse Population 3, Population 4 and Population 5 according to a certain proportion, and select the population based on the IGD-NS selection strategy to obtain Population; wherein the initial population proportion of the small-scale problem is set to 0.1, the initial population proportion of the large-scale problem is set to 0.3, and the proportion of Population 3 and Population 4 is 1:1;

[0038] Step 5: Repeat steps 1-4 above until the algorithm termination condition is reached.

[0039] On this basis, an adaptive crossover mutation operator based on the HV index is designed, mainly including:

[0040] The HV index is introduced to evaluate the performance of the population obtained by each iteration of the algorithm, and the difference ΔHV between the HV index of the current population and the HV index of the last iteration population is calculated. The crossover mutation probability value is adaptively adjusted according to its size to improve the convergence effect of the algorithm. The calculation formula of the HV index value is:

[0041]

[0042] Wherein, the larger the value of HV, the stronger the comprehensive performance of the population; c i is a hypercube formed by a non-dominated solution x and a reference point z ref as the diagonal.

[0043] Based on this, an adaptive crossover mutation operator based on the HV index is designed, denoted as:

[0044] ΔHV = HV u -HV u-1

[0045] p c = μ1 × (cos(ΔHV × π)) + τ1

[0046] p m = μ2 × (sin(ΔHV × π)) + τ2

[0047] Wherein, the cosine function is used to ensure that the probability value is within the appropriate interval; HV1 = 0, u ∈ [1, 2, …, maxcycle] is the iteration index, μ1 and μ2 are random values in the interval [0.4, 0.8] and [0.1, 0.3] respectively; τ1, τ2 are random values in the interval [0.05, 0.1], and here they are 0.1 and 0.05 respectively.

[0048] According to the present application, as further preferred, the problem knowledge driven neighborhood strategy in step S3 comprises the following sub-steps:

[0049] Step 1: Determine the maintenance time node of each machine tool under the existing scheduling scheme, and form a machine tool maintenance information set A;

[0050] Step 2: Select the maintenance machine tool in descending order based on the maintenance trigger time;

[0051] Step 3: Determine whether the machine tool should be adjusted according to the formula LBD k ≤D k ×IT k ±σ 2 ≤UBD k and If adjustment is needed, shift the subsequent affected processes on the machine tool to the left according to the greedy rule, otherwise, remove the machine tool from set A;

[0052] Step 4: Repeat the above steps until set A is empty;

[0053] Step 5: Traverse the machine tool idle position in descending order of processing time, and get the adjusted completion time based on the greedy rule;

[0054] Step 6: Adjust the tool selection based on the tooling adjustment strategy of adjacent processes.

[0055] Where σ is the diffusion coefficient, subject to normal distribution [0.0005, 0.001]; and are the initial processing time, the start time of the first process and the end time of the last process before the bth maintenance of the machine tool M k b. i,j th.

[0056] The beneficial effects of this invention are as follows: Firstly, with the goal of minimizing completion time, total energy consumption, and total production cost, a mathematical model for energy-saving batch flow scheduling in a flexible workshop, considering machine tool deterioration and preventative maintenance, is constructed. Secondly, based on the problem characteristics, a dynamic adaptive multi-objective evolutionary algorithm based on the HV index and a knowledge-driven flexible workshop batch partitioning framework are designed to solve the model. In each search stage, historical populations and randomly initialized populations are merged into the current population at a certain ratio to improve the population diversity of the algorithm. Based on this, an adaptive crossover and mutation operator based on the HV index is designed to guide the evolution of the population, improving the convergence effect of the algorithm. A knowledge-driven neighborhood operator is proposed to fully explore the local optimal solutions of the problem. This effectively addresses the impact of machine tool deterioration and preventative maintenance on multi-variety variable batch production, reduces total workshop energy consumption and production costs, shortens product completion time, and provides theoretical basis and practical support for actual production of multi-variety variable batches under dynamic and uncertain environments. Attached Figure Description

[0057] Figure 1(a) and Figure 1(b) are example diagrams of the energy-saving batch flow scheduling problem of flexible operation workshop under the machine tool deterioration and preventive maintenance environment of the present invention. Figure 1(a) is the scheduling scheme diagram before workpiece batching, and Figure 1(b) is the scheduling scheme diagram after workpiece batching optimization.

[0058] Figure 2 This is a model solution framework diagram of the hybrid method based on knowledge-driven batching method and dynamic adaptive multi-objective evolution algorithm of the present invention;

[0059] Figure 3 This is the pseudocode for the knowledge-driven neighborhood search strategy proposed in this invention;

[0060] Figures 4a-4l This is a Turkey confidence interval plot of the IGD, GD, and MS indices of each algorithm under different average maintenance times of the present invention, with a 95% confidence level.

[0061] Figures 5(1)-5(4) Box plots showing the comparison results of the knowledge-driven batching method and dynamic adaptive multi-objective evolutionary algorithm of the present invention with the HV, IGD, GD and MS indices of NSGA-II, MOPSO, ARMOEA, DNSGAII and SGEA for solving example problems.

[0062] Figure 6 This invention provides a Gantt chart illustrating the scheduling scheme for finding the optimal solution to an instance problem using the knowledge-driven batching method and the dynamic adaptive multi-objective evolutionary algorithm, as described in this embodiment.

[0063] Figure 7 This is a comparison chart of Pareto fronts for different algorithms used in example problems of this invention. Detailed Implementation

[0064] The application will be described in further detail below with reference to the drawings and embodiments, but it should be understood that the embodiments are used to explain the application and are not used to limit the application.

[0065] To make the application clearer, the method of the application is described below with a specific embodiment, as shown in Figures 1(a) and 1(b); the embodiment of the application provides an energy-saving batch flow scheduling optimization method for flexible job shop under the consideration of machine deterioration and preventive maintenance, the steps of the method are as shown in Figure 2

[0066] S1, constructing an energy-saving batch flow scheduling mathematical model for flexible job shop considering machine deterioration and preventive maintenance with the optimization objectives of completion time, total energy consumption and total production cost.

[0067] S2, model solving method based on knowledge-driven batching method and dynamic self-adaptive multi-objective evolutionary algorithm.

[0068] S3, problem knowledge-driven neighborhood search strategy.

[0069] S4, numerical experiment.

[0070] S5, example verification.

[0071] Further, the energy-saving batch flow scheduling multi-objective optimization mathematical model for flexible job shop considering machine deterioration and preventive maintenance in step S1 is:

[0072] Objective function: F = min (T e ,T m ,T c )

[0073] Wherein, F is the objective function, T e , T m , T c are the total energy consumption, completion time and total production cost of the workshop respectively.

[0074] T e = E c +E ct +E id +E m +E f

[0075]

[0076]

[0077]

[0078]

[0079] ​E f = E u × T max

[0080] where E c , E ct , E id , E m , E f and E u are the processing energy consumption, the setup energy consumption, the idle energy consumption, the handling energy consumption, the fixed energy consumption and the public energy consumption per time unit of the workshop, respectively; i, j, k and r are the indices of the workpiece, the process, the machine tool and the tooling, respectively; n', N t , q i , m are the total number of workpieces after batching, the batch size of the lth batch of workpiece J i , the number of toolings, the number of processes of workpiece J i and the number of machine tools, respectively; and P m are the standby power, the idle power of machine tool M k and the handling power of AGV, respectively; and are the clamping time of process Q i,j on machine tool M k , the tooling changeover time, the tooling adjustment time and the handling time of transferring process Q i,j' to a new machine tool for processing, respectively; X i,j,k , Y i,j,k,r , Z i,j and M k I are decision variables, X i,j takes 1 when process Q k is processed on machine tool M i,j,k , otherwise 0, Y i,j takes 1 when process Q k is processed on machine tool M r using tooling MT i,j,k,r , otherwise 0, Z i,j takes 1 when process Q i,j and the next process are processed on the same machine tool, otherwise 0, M k I takes 1 when machine tool M k is available at the scheduling time, otherwise 0.

[0081]

[0082] where T i is the end time of the q i th process of workpiece J k on machine tool M i,j,k .

[0083] The total production cost of a workpiece consists of the workpiece processing cost and the machine tool maintenance cost, expressed as:

[0084]

[0085] Among them, C i,j,k MC k and NC k They are respectively process Q i,j In machine tool M k The unit time processing cost and machine tool M k Maintenance frequency and machine tool M k The average cost of maintenance per operation.

[0086] In addition to satisfying the basic constraints of batch flow scheduling in multi-resource flexible workshops, machine tool degradation constraints must also be satisfied, expressed as:

[0087]

[0088] Among them, D k As the deterioration coefficient, IT k For machine tool M k Processed time prior to triggering maintenance For Q i,j In machine tool M k The initial processing time.

[0089] Machine tool availability is subject to time window constraints. When preventative maintenance is triggered, if there are unfinished machining tasks on the machine tool, maintenance must wait for those tasks to complete before proceeding. This is indicated as follows:

[0090]

[0091]

[0092] in, As a decision variable, if in machine tool M k Previous process Q ij For process Q gh The value is 1 for subsequent processes, otherwise it is 0; A i,j and A g,h The respective process Q is a machinable step. ij and Q gh Machine tool M k Set; W k Let M be the decision variable, and machine tool M... k If maintenance is triggered, the value is 1; otherwise, the value is 0. For machine tool M k The start time of the bth maintenance; LBD k and UPD k Machine tool Mk minimum and maximum deterioration amount of maintenance.

[0093] Machine tool maintenance constraints. Preventive maintenance can be triggered multiple times within a certain production cycle, machine M k End time point of the bth maintenance is expressed as:

[0094]

[0095] Wherein, MT k is the machine M k Time required for maintenance.

[0096] According to the application, as further preferred, the step S2 comprises the following sub-steps:

[0097] Global optimization is carried out based on MFO operator, and the searched formula is: x u =dis(x u ,f u )×e bt ×cos(2πt)+f u

[0098] Wherein, dis(x u ,f u ) represents the Euclidean distance between the moth x u And the flame f u .

[0099] b is a defined spiral shape constant, taking 0.01.

[0100] t is the path coefficient, taking the value range [-1, 1].

[0101] A population diversity preservation strategy based on historical population is designed. By fusing the historical population into the current population in a certain proportion in each search stage, the diversity of the current population is improved. The specific operation is as follows:

[0102] Step 1: use MFO operator to optimize the initialized population Population to get the offspring population Population2;

[0103] Step 2: fuse Population and Population2 in a 1:1 ratio, and select the merged population based on IGD-NS selection strategy to get Population3;

[0104] Step 3: optimize the population Population3 based on the synergistic evolution operator to get the offspring population Population4;

[0105] Step 4: Randomly initialize a population to obtain Population 5, fuse Population 3, Population 4 and Population 5 according to a certain proportion, and select the population based on the IGD-NS selection strategy to obtain Population; wherein the proportion of the initial population of the small-scale problem is set to 0.1, the proportion of the initial population of the large-scale problem is set to 0.3, and the proportion of Population 3 and Population 4 is 1:1;

[0106] Step 5: Repeat steps 1-4 above until the algorithm termination condition is reached.

[0107] On this basis, an adaptive crossover mutation operator based on the HV index is designed, mainly including:

[0108] The HV index is introduced to evaluate the performance of the population obtained by each iteration of the algorithm, and the difference ΔHV between the HV index of the current population and the HV index of the last iteration population is calculated. The crossover mutation probability value is adaptively adjusted according to its size to improve the convergence effect of the algorithm. The calculation formula of the HV index value is:

[0109]

[0110] Wherein, the larger the value of HV, the stronger the comprehensive performance of the population; c i is a hypercube formed by a non-dominated solution x and a reference point z ref as the diagonal.

[0111] Based on this, an adaptive crossover mutation operator based on the HV index is designed, denoted as:

[0112] ΔHV = HV u -HV u-1

[0113] p c = μ1 × (cos(ΔHV × π)) + τ1

[0114] p m = μ2 × (sin(ΔHV × π)) + τ2

[0115] Wherein, the cosine function is used to ensure that the probability value is within the appropriate interval; HV1 = 0, u ∈ [1, 2, …, maxcycle] is the iteration index, μ1 and μ2 are random values in the interval [0.4, 0.8] and [0.1, 0.3] respectively; τ1, τ2 are random values in the interval [0.05, 0.1], and here they are 0.1 and 0.05 respectively.

[0116] Further, the pseudo code of the problem knowledge driven neighborhood strategy in step S3 is as follows Figure 3As shown, the specific implementation steps are:

[0117] Step 1: Determine the maintenance time node of each machine tool under the existing scheduling scheme, and form a machine tool maintenance information set A;

[0118] Step 2: Select the maintenance machine tool in descending order based on the maintenance trigger time;

[0119] Step 3: According to the formula LBD k ≤D k ×IT k ±σ 2 ≤UBD k And Determine whether the machine tool should be adjusted. If adjustment is needed, shift the subsequent affected processes on the machine tool to the left according to the greedy rule, otherwise, remove the machine tool from set A;

[0120] Step 4: Repeat the above steps until set A is empty;

[0121] Step 5: Traverse the machine tool idle position in descending order of processing time, and get the adjusted completion time based on the greedy rule;

[0122] Step 6: Adjust the tool selection based on the tooling adjustment strategy of adjacent processes.

[0123] Where σ is the diffusion coefficient, which follows a normal distribution [0.0005, 0.001]; And are the initial processing time of the first process Q k of the machine tool M i,j after the bth maintenance, the start time of the first process, and the end time of the last process before the bth maintenance.

[0124] Further, the step S4 includes the following sub-steps:

[0125] To verify the effectiveness of the flexible job-shop energy-saving batch flow scheduling optimization method proposed in the present application, 90 test cases are randomly generated, and the proposed knowledge-driven batching method and dynamic self-adaptive multi-objective evolutionary algorithm hybrid method, NSGA-II, MOPSO, ARMOEA, DNSGAII and SGEA are used for solving; these instances include 10-20 workpieces, 10-15 machine tools, an average number of processes of each workpiece of 6-15, an average of 3 machine tools that can be selected for processing per process, an average batch size of 10-30, and a unit maintenance cost of the machine tool obeying a normal distribution[3, 6], and an average maintenance time of the machine tool of 10-50. The statistical results of 10 times of solving of each algorithm are shown in Table 1. As can be seen from Table 1, for the average values of each index, the proposed knowledge-driven batching method and dynamic self-adaptive multi-objective evolutionary algorithm hybrid method are significantly better than the comparative algorithms. The Turkey confidence interval chart of the IGD, GD and MS indexes of each algorithm under different average maintenance times with a confidence level of 95% is shown in Figures 4a-4l

[0126] Table 1 Statistical results of different algorithms on test cases (average value)

[0127]

[0128] Further, the instance in step S5 is verified from a production workshop of a machine tool manufacturing enterprise, which has 16 machining equipment such as gantry machining centers, horizontal numerical control machine tools, medium-sized numerical control machine tools, small-sized numerical control machine tools, vertical machining centers, boring and milling machining centers, etc. in a certain workshop of the enterprise, the processing batch sizes of the bed body, column and slide parts of an order batch are 10, 15 and 20 respectively, the process numbers of each part are 22, 15 and 18 respectively, and there are 9 toolings that can be used. The deterioration coefficient of the machine tool is set to obey a uniform distribution [0.005, 0.15] according to the production experience and historical data information of workers, the diffusion coefficient obeys a uniform distribution [0.0005, 0.015], the average maintenance time of the machine tool is 30 min, and the maintenance costs of the machine tools are 10, 15, 20, 10, 15, 20, 10, 15, 20, 10, 15, 20, 10, 15, 20, 10, 15, 20 and 15 respectively. The proposed knowledge-driven batching method and dynamic self-adaptive multi-objective evolutionary algorithm hybrid method, NSGA-II, MOPSO, ARMOEA, DNSGAII and SGEA are used for solving, and the comparison results of each index of 10 times of solving are shown in Figures 5(1)-5(4) Figures 5(1)-5(4) ​​It can be seen that the proposed knowledge-driven batch method and improved multi-objective evolutionary algorithm hybrid method are significantly better than the comparative algorithms in each index. The proposed method improves the average of HV, IGD, GD and MS by 1.2%, 56.7%, 53.1% and 39.4%, respectively, which can effectively reduce the total energy consumption and cost of the workshop and shorten the product completion time. In addition, DAMOEA has significant differences with NSGA-II, MOPSO, ARMOEA, DNSGAII and SGEA in all evaluation indexes, which further verifies the effectiveness of the proposed optimization method. The optimal scheduling scheme is shown in Fig. 2344 Figure 6 The number of sub-batches of each workpiece is 2, 3 and 5, respectively, and the sub-batch quantity is 4, 6, 4, 4, 4 and 4, respectively. The completion time is 3960.29 min, the total energy consumption of the workshop is 3066.12 kW·min, and the total processing cost is 76100.9 CNY. The Pareto front of different algorithms in the actual case is shown in Fig. Figure 7 It can be seen that the number of Pareto solutions obtained based on DAMOEA is more and the distribution is better, which is conducive to the decision maker to select a satisfactory scheduling scheme according to the actual production.

[0129] The above examples only illustrate the technical concept and characteristics of the present application, the purpose is to enable those skilled in the art to understand the content of the present application and to implement it, and cannot limit the protection scope of the present application. Any equivalent changes or modifications made according to the essence of the present application shall fall within the protection scope of the present application.

Claims

1. A flexible job-shop energy-saving batch flow scheduling optimization method considering machine deterioration and preventive maintenance, characterized in that, The method comprises the following steps: S1, constructing a flexible job shop energy-saving batch flow scheduling mathematical model considering machine tool deterioration and preventive maintenance with the optimization target of completion time, total energy consumption and total production cost; S2, a knowledge-driven batch method and a dynamic self-adaptive multi-objective evolutionary algorithm are used to solve the flexible job shop energy-saving batch flow scheduling mathematical model considering machine tool deterioration and preventive maintenance, a population diversity introduction strategy based on historical population is proposed to enhance the algorithm's ability to respond to environmental changes, an adaptive crossover and mutation operator based on hyper-volume index is developed to balance the diversity and convergence of the algorithm, and a problem knowledge-driven neighborhood search strategy is proposed to further explore the local optimal solution of the flexible job shop energy-saving batch flow scheduling problem considering machine tool deterioration and preventive maintenance; The population diversity maintenance strategy based on historical population comprises the following steps: The historical population is fused into the current population in a certain proportion in each search stage to improve the diversity of the current population, and the specific operation is as follows: Step 1: the MFO operator is used to optimize the initialized population Population to obtain a child population Population2; Step 2: Population and Population2 are fused in a 1:1 ratio, and Population3 is obtained by selecting the merged population based on the IGD-NS selection strategy; Step 3: the cooperative evolution operator is used to optimize the population Population3 to obtain a child population Population4; Step 4: the population is randomly initialized to obtain Population5, Population3, Population4 and Population5 are fused in a certain proportion, and the population is selected based on the IGD-NS selection strategy to obtain Population; wherein the initial population proportion of the small-scale problem is set to 0.1, the initial population proportion of the large-scale problem is set to 0.3, and the proportion of Population3 and Population4 is 1:1; Step 5: repeat steps 1-4 until the algorithm termination condition is reached; The adaptive crossover and mutation operator based on the HV index guides the evolution strategy of the population as follows: The HV index is introduced to evaluate the performance of the population obtained by each iteration of the algorithm, the difference ΔHV between the HV index of the current population and the HV index of the population in the last iteration is calculated, the crossover and mutation probability values are adaptively adjusted according to the size, and the convergence effect of the algorithm is improved; wherein the calculation formula of the HV index value is: Wherein, the greater the value of HV indicates the stronger the comprehensive performance of the population; c i is a hypercube formed by a certain non-dominated solution and a reference point z ref as the diagonal. Based on this, the adaptive crossover and mutation operator based on the HV index is designed, which is represented as: AHV = HV u -HV u-1 p c = μ1 x (cos(ΔHV x π)) + τ1 p m = μ2 x (sin(ΔHV x π)) + τ2 wherein, AHV is the difference between the current population and the last iteration population HV index; p c and p m are the crossover probability and mutation probability, respectively; the cosine function is used to ensure that the probability value is in the appropriate interval; HV1=0, u∈[1, 2, …, maxcycle] is the iteration index, and maxcycle is the maximum number of iterations of the algorithm; HV u is the HV index value of the current population at the u iteration; HV u 1 is the HV index value of the current population at the u-1 iteration; μ1 and μ2 are random values in the intervals [0.4, 0.8] and [0.1, 0.3], respectively; τ1, τ2 are random values in the interval [0.05, 0.1], and herein, 0.1 and 0.05 are taken, respectively; The knowledge-driven neighborhood operator comprises the following steps: Step 1: determine the maintenance time node of each machine tool under the existing scheduling scheme to form a machine tool maintenance information set A; Step 2: select the maintenance machine tool in descending order of maintenance trigger time; Step 3: LBD according to formula k ≤D k ×IT k ±σ2≤UPD k and determining whether the machine tool should be adjusted, if adjustment is needed, then left shifting the machine tool according to the greedy rule for the subsequent affected processes, otherwise eliminating the machine tool from the set A; Wherein, σ is the diffusion coefficient, which is normally distributed [0.0005, 0.001]; and is the initial processing time of the machine tool M k is the start time of the first process and the end time of the last process before the bth maintenance; MT i,j is the initial processing time of the machine tool M k is the time required for maintenance of the machine tool M k is the time required for maintenance of the machine tool M Step 4: repeat the above steps until the set A is empty; Step 5: traverse the idle position of the machine tool in descending order of processing time, and obtain the adjusted completion time based on the greedy rule; Step 6: Adjust the tool selection based on the tooling post-adjustment strategy of the previous step.

2. The flexible job-shop energy batch-flow scheduling optimization method considering machine deterioration and preventive maintenance according to claim 1, wherein, The multi-objective optimization mathematical model of energy-saving batch flow scheduling in flexible job shop considering machine deterioration and preventive maintenance includes: The optimization objective of the model is to minimize the makespan, total energy consumption, and total processing cost, and the calculation formula is: T e = E c + E ct + E id + E m + E f E f = E u x T max Among them, T e With T max These are the total energy consumption and completion time of the workshop, respectively; E c E ct E id E m E f and E u These represent processing energy consumption, processing preparation energy consumption, idle energy consumption, transportation energy consumption, stationary energy consumption, and common energy consumption generated per unit time in the workshop; i, j, k, and r are the indices of the workpiece, process, machine tool, and tooling, respectively; n', N t q i m represents the total number of workpieces after batching, and J represents the number of workpieces. i Batch size, tooling quantity, and workpiece J of the lth sub-batch i The number of processes and the number of machine tools; P i,j,k , and P m Machine tool M k Processing step Q i,j Average power, standby power, idle power, and AGV handling power; t i,j,k , and They are respectively process Q i,j The machining time, clamping time, tooling change time, tooling adjustment time, and process transfer Q required for machine tool processing. i,j' The time required for transporting materials to the new machine tool and the idle time required for processing; X i,j,k Y i,j,k,r Z i,j and MI k As a decision variable, when process Q i,j In machine tool M k X during processing i,j,k Set to 1, otherwise set to 0. If process Q i,j In machine tool M k Upper machining using tooling MT r Time Y i,j,k,r Set to 1, otherwise set to 0. If process Q i,j If it is processed on the same machine tool as the next process, then Z i,j Set the value to 1, otherwise set it to 0. If machine tool M k If MI is available during the scheduling time, then k Select 1 if the value is 1, otherwise select 0. wherein, the end time of the qth process on the machine tool M i of the workpiece J i k the end time of the qth process on the machine tool M​ where T c is the total production cost; C i,j,k , MC k and NC k are the processing cost per unit time of the process Q i,j on the machine M k , the number of maintenance of the machine M k and the average cost of one maintenance of the machine M k , respectively. In addition to meeting the basic constraint conditions of multi-resource flexible job shop batch flow scheduling, the machine deterioration constraint must also be met, which is represented as: wherein Q is i,j the initial machining time on the machine tool M k ; Machine available time window constraint: When preventive maintenance is triggered, if there is a processing task on the machine that has not been completed, it needs to wait for its completion before maintenance, which is represented as: W k ∈{0,1},k∈[1,m],if LBD k ≤D k ×ET i,j,k ≤UPD k ,W k Randomly generated within the optional set; If D k x ET i,j,k ≥ UPD k , W k = 1; otherwise, W k = 0 where, For the decision variable, if the machine tool M k Q ij is the successor of Q gh ; A i,j and A g,h are the sets of machine tools M ij that can process the workpieces Q gh and Q k , respectively; W k is a decision variable that takes the value 1 if a machine tool M k triggered a maintenance, and 0 otherwise; is the starting point in time of the b-th maintenance of a machine tool M k . Machine maintenance constraint: preventive maintenance can be triggered several times within a production cycle, machine M k End time point of the bth maintenance is expressed as:

Images