Flexible job shop scheduling and maintenance decision joint optimization method considering working conditions
By building a time-varying equipment reliability model and two-stage joint optimization method, the problems of equipment failure and delivery delay in flexible operation workshops are solved, and equipment failure-free operation and production stability are achieved, and shutdown losses and delivery are reduced. Delay.
Patent Information
- Application Number
- CN202411111504.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-14
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-08-14
AI Technical Summary
The existing research on scheduling and maintenance decisions in flexible operation workshops failed to fully consider the actual operating performance of the equipment and the coupling between production tasks and maintenance activities, resulting in equipment failure and downtime losses and delivery delays.
Build a equipment reliability model under time-varying conditions, combine maintenance cost rate and availability indicators, design a two-stage joint optimization method for production scheduling and maintenance decision-making, and use mixed discrete particle swarms and variable neighborhood search methods to generate an initial production plan. And dynamically optimize production and maintenance plans to achieve failure-free operation of equipment.
It significantly reduces downtime losses and delivery delays caused by equipment failures, balances equipment utilization, and formulates cost-effective production and maintenance plans to ensure the equipment's fault-free operation and uninterrupted production.
Smart Images

Figure CN118941046A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a technology in the field of intelligent manufacturing, specifically a joint optimization method for flexible job shop scheduling and maintenance decision-making considering working conditions. Background Art
[0002] Flexible manufacturing systems are widely used in industries such as automobiles, aerospace, and semiconductors. Flexible job shop scheduling enables mixed production of multiple types of products by rationally allocating system resources. However, the equipment exhibits multi-stage nonlinear failure characteristics under the processing conditions of different types of products, which aggravates the random uncertainty of equipment failures and thus affects the delivery time of products. Therefore, the joint optimization of production scheduling and maintenance decisions is the key to balancing system production efficiency and equipment maintenance needs. At present, research on flexible job shop scheduling and maintenance decisions mainly focuses on optimizing production scheduling results using scheduled maintenance plans, which has the following shortcomings: it does not take into account the actual operating performance of the equipment and cannot respond to the uncertainty of equipment maintenance needs under variable working conditions; in addition, it ignores the coupling between production tasks and maintenance activities, and the static joint optimization scheme cannot achieve global economic optimality. Summary of the invention
[0003] In view of the above-mentioned deficiencies in the prior art, the present invention proposes a joint optimization method for flexible job shop scheduling and maintenance decision-making considering working conditions. An equipment reliability model under time-varying working conditions is constructed, and the maintenance cost rate and equipment availability index are considered to realize equipment maintenance planning, and a two-stage joint optimization method for production scheduling and maintenance decision-making is designed. The present invention can greatly reduce the downtime losses caused by equipment failure, alleviate customer dissatisfaction with delayed delivery, and balance the utilization of multiple devices. By formulating cost-effective production and maintenance plans, it ensures trouble-free operation of equipment and uninterrupted smooth production.
[0004] The present invention is achieved through the following technical solutions:
[0005] like Figure 1 As shown, the present invention relates to a joint optimization method for flexible job shop scheduling and maintenance decision-making considering working conditions. In the first stage, a production scheduling model is constructed according to the production capacity of workshop equipment, product process routes and order task requirements, and an initial production plan is generated based on a hybrid discrete particle swarm and a variable neighborhood search method; in the second stage, based on the maintenance decision model, the equipment maintenance requirements under the production plan are analyzed, and a maintenance plan is formulated. When the production tasks and maintenance activities on the same equipment conflict, the production plan is dynamically optimized based on the variable neighborhood search method and the equipment status is updated in real time, and an economically optimized production plan and maintenance plan are simultaneously formulated.
[0006] The production scheduling model aims to minimize the penalty cost of order delay and the penalty cost of load balancing between equipments, and establishes an integer programming model to obtain the optimal initial production scheduling plan, which specifically includes:
[0007] Step 1: Define the parameters as follows: M = {M1, M2, …, M K} is the workshop equipment set, index k; P = {P1,P2, …, P I} is a product collection, indexed by i; i = {O i1 , O i2 , …, } is product P i The process set of M is indexed as j; ij For process O i Optional equipment set, process O ij Processing time PT ijk Depends on the selected device M K ; Z = {z1, z2, …, z K} is the number of processing positions of the equipment, indexed by z; D i For product P i Delivery period, For product P i Unit time penalty for delayed delivery; C b Penalty for load balancing between devices;
[0008] Step 2: Define the decision variables as follows: ijkz ∈{0,1},x ijkz =1 indicates process O ij Processing at the zth position of equipment k, otherwise = 0; >0 indicates process O ij Start processing time; >0 indicates process O ij completion time;
[0009] Step 3: Establish a model optimization goal to minimize the total cost. The total cost is composed of the total delay penalty cost and the equipment load balancing penalty cost. The total delay penalty cost refers to the sum of the delayed delivery penalties of all products in the order task. Indicates that TTC i is the delay penalty of product i. If the delivery time of the product is later than the required delivery time, a corresponding penalty cost is generated, which is proportional to the delay time. Otherwise, the delay penalty is 0. The device load balancing penalty cost is proportional to the load balancing level of all devices in the system, expressed as TBC. The smaller the variance of all device loads, the smaller the cost item. The optimization objective and the total cost of each item are expressed as: ,in , , ;
[0010] Step 4: Establish constraints to ensure that each process of the product can only be assigned to one position on one piece of equipment for processing, expressed as: ;
[0011] Step 5: Establish constraints to ensure that the processing of product processes strictly follows the process flow, that is, the start time of a process must not be earlier than the completion time of the previous process of the product, expressed as ;
[0012] Step 6: Establish constraints to ensure that once a process starts, it cannot be interrupted before completion, expressed as ;
[0013] Step 7: Establish constraints to ensure that device M k Each processing position can only be used to process one product at most, expressed as ;
[0014] Step 8: Establish constraints to ensure that the processing position of each device is assigned to each process of the product in sequence, expressed as ;
[0015] Step 9: Establish constraints to ensure that there is no overlap between processes assigned to the same equipment, that is, the start time of the current processing position of each equipment must not be earlier than the completion time of the previous processing position, expressed as ;
[0016] Step 10: Based on the above parameters, decision variables, optimization objectives and constraints, the flexible job shop scheduling model is established as follows: ,in , ,in ; The first constraint of the model is , the second constraint is ,in , the third constraint is , the fourth constraint is , the fifth constraint is .
[0017] The hybrid discrete particle swarm and variable neighborhood search method comprises the following steps:
[0018] Step a, initialization parameters, including particle swarm size, maximum number of iterations, crossover probability and mutation probability;
[0019] Step b, a two-layer coding method is used to represent a particle or a feasible solution, wherein the two layers of coding are process sequence and equipment sequence, which are denoted as OS and MS respectively; the elements in OS represent the index of the processed product, and the number of times the index appears represents the number of processes of the corresponding product; the elements in MS sequence represent the processing equipment selected for each process in OS;
[0020] Step c: Optimize the randomly generated initial particles by discrete particle swarm method, and update the particles with the objective function as the fitness value. The process includes three parts. The first part randomly selects the OS sequence or MS sequence of the particle for mutation operation. For the OS sequence, randomly select an element at one position and insert it into other positions to change the order of the OS sequence elements. For the MS sequence, randomly select multiple positions to change the processing equipment selection of the corresponding process, and generate new particles represented as ;
[0021] Step d, the second part randomly selects an individual optimal particle in the crossover learning iteration process The first crossover operation is: randomly divide the order task P into two non-empty subsets P 1 and P 2 , from the current particle Take out the 1 The products in the collection are copied to the new particles, maintaining the element information of the process sequence and equipment sequence; Remove the 1 The product information in the original order is copied to the new particles. ; The second crossover operation is: randomly select the first S positions of the particle encoding sequence, and Take out the products in the first S positions in turn and copy them to the new particles, keeping the element information of the process sequence and equipment sequence; Remove the particles contained in the current The product information in the first S positions, copy the remaining products to the new particles in the original order ;
[0022] Step e, the third part randomly selects a global optimal particle in the crossover learning iteration process The first crossover operation is: randomly divide the order task P into two non-empty subsets P 1 and P 2 , from the current particle Take out the 1 The products in the collection are copied to the new particles, keeping the element information of the process sequence and equipment sequence; in the global optimal particle Remove the 1The product information in the original order is copied to the new particles. ; The second crossover operation is: randomly select the first S positions of the particle encoding sequence, and Take out the products in the first S positions in turn and copy them to the new particles, keeping the element information of the process sequence and equipment sequence; in the global optimal particle Remove the particles contained in the current The product information in the first S positions, copy the remaining products to the new particles in the original order ;
[0023] Step f, when the iteration termination condition is reached, the particles with the top 10% of fitness values are selected as the input of the variable neighborhood search method;
[0024] Step g: In the variable neighborhood search method, a deep local search is performed on the current particle through mutation and crossover operations. For the mutation operation, an element in the MS sequence is randomly selected to change the processing equipment selection of the corresponding process; for the crossover operation, a position in the OS sequence is randomly selected to insert the element at the position into other positions;
[0025] Step h: When the iteration termination condition is reached, the particle with the best fitness value is output.
[0026] The maintenance decision model considers the evolution of equipment failure rate under time-varying working conditions, combines the maintenance parameters of each equipment, and constructs a multi-objective optimization equipment maintenance decision model. The model is solved to obtain the optimal maintenance time interval of each maintenance cycle of the equipment, which specifically includes:
[0027] Step ①: Define the parameters as follows: B={r1, r2, …, r K} represents the set of equipment maintenance cycles, the preventive cycle index of equipment k is r; the duration of a preventive maintenance activity of equipment k is expressed as , the corresponding preventive maintenance activity cost is ; When equipment k fails, the duration of corrective maintenance activities is expressed as , the corresponding corrective maintenance activity cost is ; The number of failures before the zth processing position of equipment k is recorded as u kz ; W is a very large positive integer;
[0028] Step ②: Define the decision variables as follows: y rkz ∈{0,1},y rkz =1 means that the preventive maintenance activity is performed after the zth processing position in the rth maintenance cycle of equipment k, otherwise y rkz =0; >0 indicates the actual preventive maintenance time of equipment k in the rth maintenance cycle;
[0029] Step 3: Define the intermediate variables as follows: rk (t|X) represents the failure rate function of equipment k in the rth maintenance cycle; represents the optimal maintenance time interval of equipment k in the rth maintenance cycle, represents the actual maintenance time interval of device k in the rth maintenance cycle, represents the optimal maintenance operation time point of equipment k in the rth maintenance cycle, represents the optimal maintenance operation end time point of equipment k in the rth maintenance cycle, Indicates the actual maintenance operation end time of equipment k in the rth maintenance cycle;
[0030] Step ④: Establish equipment reliability model under time-varying working conditions, λ rk (t|X) =αλ k (t)f(βX), where X = [X1, X2, …, X f , …, X F ] indicates the impact on equipment M during processing k A set of factors for the failure rate, β = [β1, β2,…, β F ] represents the coefficient of the corresponding factor in X;
[0031] Step ⑤: Equipment M k The baseline failure rate function is represented by Weibull distribution λ k (t)= , where θ k and η k They represent the shape parameter and scale parameter in the Weibull distribution of device k, respectively;
[0032] Step ⑥: Build equipment maintenance planning model V with the goal of minimizing maintenance cost rate and maximizing equipment availability rk = By solving the derivative equation of the model and making the derivative equation equal to 0, the optimal maintenance time interval of the equipment cycle can be obtained. ; Among them, the maintenance cost rate is defined as , equipment availability is defined as = , by respectively making d / dT rk = 0 and d / dT rk =0, the minimum maintenance cost rate value can be obtained and the maximum equipment availability value ;
[0033] Step 7: Determine the time period for equipment k to perform maintenance activities in the rth cycle according to the optimal maintenance time interval of the equipment. , ], when the period is in the standby period of device k, the actual start time of the maintenance activity of device k in the rth cycle is = On the contrary, when [ , ] When the normal processing process of the process at the zth position of equipment k conflicts, the cost balance of preventive maintenance activities and the cost balance of corrective maintenance activities are constructed, and the actual maintenance time of the equipment is optimized based on the cost balance selection method;
[0034] Step ⑧: When the preventive maintenance activity of equipment k in the rth cycle is scheduled to be implemented before the zth position starts processing, the start time of the maintenance activity is denoted as t a , when satisfied When t a Expressed as , when not satisfied When t a Expressed as ; The cost balance of corrective maintenance activities is expressed as , expressed as , which is equivalent to ; The cost balance of preventive maintenance activities is expressed as , expressed as ;
[0035] Step ⑨: When the preventive maintenance work of equipment k in the rth cycle is scheduled to be carried out after the completion of the zth position, the start time of the maintenance activity is denoted as t d , expressed as ; The cost balance of corrective maintenance activities is expressed as , expressed as , which is equivalent to ; The cost balance of preventive maintenance activities is expressed as , expressed as ;
[0036] Step ⑩: The cost surplus optimization method described above combines the economic benefits of advancing or delaying the equipment maintenance time to determine the actual maintenance activity time of each equipment. , the decision variable y rk(z-1) =1, =t a On the contrary, y rkz =1, =t d .
[0037] The production scheduling and maintenance decision joint optimization model is to minimize the total delay penalty cost, equipment load balancing penalty cost, and total maintenance cost, where the total delay penalty cost and device load balancing penalty costs The same expression as in the first-stage production scheduling model; the total maintenance cost is expressed as , TMC k is the maintenance cost of equipment k; the optimization target and the total maintenance cost are expressed as: ,in TMC k = ∑ r ∊ B ∑ z ∊ Z y rk [ C k p + C k f × ∫ 0 T rk ' λ rk ( t | X ) dt ] ; ∀ k ∊ M ; Specifically include:
[0038] Step 1: Establish constraints to ensure that each device M k At most one maintenance activity can be performed on the system, expressed as: ;
[0039] Step 2: Establish constraints to ensure that the maintenance cycle of each device is arranged in sequence, expressed as ;
[0040] Step 3: Establish constraints to ensure that the actual start time of the equipment maintenance activity is later than the completion time of the current processing location process, expressed as ;
[0041] Step 4: Establish constraints to ensure that the start time of the next position of the equipment is later than the end time of the maintenance activity at the current position. , should meet the start processing time of the next position of the equipment Greater than or equal to the end time of the maintenance activity at the current location ;
[0042] Step 5: Based on the above parameters, decision variables, optimization objectives, constraints and the constraints in the first stage production scheduling model, the second stage production scheduling and maintenance decision joint optimization model is established as follows: The overall goal is expressed as ,in TMC k = ∑ r ∊ B ∑ z ∊ Z y rk [ C k p + C k f × ∫ 0 T rk ' λ rk ( t | X ) dt ] ; ∀ k ∊ M ,in ; The first constraint is , the second constraint is , the third constraint is , the fourth constraint is .in P= ( 1- ∑ r ∊ B y rk ) [ ∑ i ' ∊ P ∑ j ' ∊ O i ' x i ' j ' kz t i ' j ' e + u k ( z +1) T k f ] , .
[0043] Technical Effects
[0044] Aiming at the fact that the failure behavior of equipment dependent on production and the strong coupling between production tasks and maintenance activities are not fully considered in the traditional flexible job shop scheduling and maintenance decision research, this paper proposes a joint optimization method for flexible job shop scheduling and maintenance decision taking into account working conditions. Specifically: construct an equipment reliability model considering time-varying working conditions, propose an equipment maintenance planning model based on maintenance cost rate and availability index to solve the optimal maintenance time interval for each equipment, and propose a maintenance decision model based on cost surplus optimization in view of the non-parallelism of production tasks and maintenance activities for the same equipment; for the first time, establish an integer programming model with the minimization of total cost including total delay penalty cost, load balancing penalty cost between equipment and total maintenance cost as the optimization goal, and systematically consider the equipment processing capacity, order task requirements, product process route, and production-maintenance coupling constraints of the flexible job shop; innovatively construct a two-stage joint optimization framework for flexible job shop scheduling and maintenance decision-making, and obtain a highly economical and operational production and maintenance plan through a hybrid discrete particle swarm and variable neighborhood search method, so that the total production and maintenance cost of the flexible job shop including the total delay penalty cost, load balancing penalty cost between equipment and total maintenance cost is significantly reduced compared with the traditional optimization method, which provides a high reference guidance value for manufacturing enterprises to formulate and optimize the production and maintenance plans of flexible job shops in real time. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] Figure 1 It is a flow chart of the present invention;
[0046] Figure 2 This is a Gantt chart of the initial production plan obtained based on the hybrid discrete particle swarm and variable neighborhood search method in the first stage of an embodiment of the present invention;
[0047] Figure 3 It is a diagram of the equipment failure rate evolution whether the working condition is considered under the initial production plan in the embodiment of the present invention;
[0048] Figure 4 A comparison diagram of the optimal and actual maintenance activity execution periods of the equipment in an embodiment of the present invention;
[0049] Figure 5 A Gantt chart of a flexible job shop production and maintenance plan obtained after optimization based on a variable neighborhood search method in an embodiment of the present invention; DETAILED DESCRIPTION
[0050] This embodiment relates to a method for joint optimization of flexible job shop scheduling and maintenance decision-making considering working conditions, including the following steps:
[0051] Step 1: Design decision variables and optimization objectives according to the production scheduling problem requirements of the flexible job shop: The present invention optimizes the total cost including the total delay penalty cost and the load balancing penalty cost between equipment by rationally planning the production plan of the flexible job shop order tasks. Therefore, the decision variables and optimization objectives are designed as follows:
[0052] (1-1) The decision variables include the decision variables related to the order production plan, x ijkz is a binary decision variable that determines whether process j of product i is processed on equipment k. If x ijkz =1 means yes, otherwise x ijkz =0; Decision-making: the start time of process j for product i; Determine the completion time of process j for product i.
[0053] (1-2) The optimization goal is to minimize the total delay penalty cost , and the total cost of the sum of the device load balancing penalty cost TBC, where TTC i is the delay penalty of product i. If the delivery time of the product is later than the required delivery time, a corresponding penalty cost is generated, which is proportional to the delay time. Otherwise, the delay penalty is 0. TBC is proportional to the load balance level of all equipment in the workshop. The smaller the variance of the equipment load, the smaller the cost item. Therefore, the optimization objective of the problem and the cost design of each part are as follows: ,in , ,in ; Among them, M, P, O iThey represent the equipment set in the workshop, the order task product set and the process set of product i, k is the equipment serial number, i is the order task product serial number, j is the product process serial number, M ij Indicates process O ij Optional equipment set, PT ijk Indicates process O ij On device M K The time required for processing; Z represents the number of processing positions of workshop equipment, D i Indicates product P i Delivery period, For product P i The unit time penalty for delayed delivery, C b It represents the penalty of load balancing between devices. The above parameters are obtained according to the actual production situation of the workshop;
[0054] Step 2: According to the actual production situation, construct the constraints of the model as follows:
[0055] (2-1) Constraints are established to ensure that each process of the product can only be assigned to one position on one piece of equipment for processing, expressed as: ;
[0056] (2-2) Constraints are established to ensure that the processing of product processes strictly follows the process flow, that is, the start time of a process must not be earlier than the completion time of the previous process of the product, expressed as ;
[0057] (2-3) Constraints are established to ensure that once a process starts, it cannot be interrupted before completion, expressed as ;
[0058] (2-4) Establish constraints to ensure that device M k Each processing position can only be used to process one product at most, expressed as ;
[0059] (2-5) Constraints are established to ensure that the processing position of each device is assigned to each process of the product in sequence, expressed as ;
[0060] (2-6) Constraints are established to ensure that there is no overlap between processes assigned to the same equipment, that is, the start time of the current processing position of each equipment must not be earlier than the completion time of the previous processing position, expressed as ;
[0061] Step 3: Use the hybrid discrete particle swarm and variable neighborhood search method to solve the integer programming model with the optimization goal of minimizing the total cost, and generate the initial production plan before the workshop starts processing, including:
[0062] (3-1) Set the particle swarm size, maximum number of iterations, crossover probability and mutation probability parameters;
[0063] (3-2) A two-layer coding method is used to represent a particle or a feasible solution, which includes a process sequence and an equipment sequence, which are represented as OS and MS respectively. The elements in OS represent the index of the processed product, and the number of times the index appears represents the number of processes of the corresponding product. The elements in MS sequence represent the processing equipment selected for each process in OS.
[0064] (3-3) The objective function is used as the fitness value to update the particle, and the OS sequence or MS sequence of the particle is randomly selected for mutation operation. For the OS sequence, an element at a position is randomly selected and inserted into another position to change the order of the OS sequence elements; for the MS sequence, multiple positions are randomly selected to change the processing equipment selection of the corresponding process, and the new particle is generated as ;
[0065] (3-4) Randomly select an individual optimal particle in the crossover learning iteration process The first crossover operation is to randomly divide the order task P into two non-empty subsets P 1 and P 2 , from the current particle Take out the 1 The products in the collection are copied to the new particles, maintaining the element information of the process sequence and equipment sequence; Remove the 1 The product information in the original order is copied to the new particles. The second crossover operation is: randomly select the first S positions of the particle encoding sequence, and Take out the products in the first S positions in turn and copy them to the new particles, keeping the element information of the process sequence and equipment sequence; Remove the particles contained in the current The product information in the first S positions, copy the remaining products to the new particles in the original order ;
[0066] (3-5) Randomly select a global optimal particle in the crossover learning iteration process The first crossover operation is to randomly divide the order task P into two non-empty subsets P 1 and P 2 , from the current particle Take out the 1 The products in the collection are copied to the new particles, keeping the element information of the process sequence and equipment sequence; in the global optimal particle Remove the1 The product information in the original order is copied to the new particles. The second crossover operation is: randomly select the first S positions of the particle encoding sequence, and Take out the products in the first S positions in turn and copy them to the new particles, keeping the element information of the process sequence and equipment sequence; in the global optimal particle Remove the particles contained in the current The product information in the first S positions, copy the remaining products to the new particles in the original order .
[0067] (3-6) When the iteration termination condition is reached, the particles with the top 10% of fitness values are selected as the input of the variable neighborhood search method;
[0068] (3-7) In the variable neighborhood search method, a deep local search is performed on the current particle through mutation and crossover operations. For the mutation operation, an element in the MS sequence is randomly selected to change the processing equipment selection of the corresponding process; for the crossover operation, a position in the OS sequence is randomly selected and the element at that position is inserted into other positions;
[0069] (3-8) When the iteration termination condition is reached, the particle with the best fitness value is output as the initial production scheduling plan.
[0070] Step 4: Design decision variables and optimization objectives according to the requirements of the joint scheduling problem of flexible job shop scheduling and equipment maintenance decision: This invention optimizes the total cost including the total delay penalty cost, the load balancing penalty cost between equipment and the total maintenance cost by considering the time-varying working conditions and the strong coupling effect between production tasks and maintenance activities, formulates the maintenance plan and optimizes the initial production plan of the first stage at the same time, so the design decision variables and optimization objectives are as follows:
[0071] (4-1) Set the decision variable: y rkz is a binary decision variable for deciding whether to perform preventive maintenance activities after the zth processing position in the rth maintenance cycle of equipment k. If y rkz =1 means yes, otherwise y rkz =0; The actual preventive maintenance time point of decision-making equipment k in the rth maintenance cycle;
[0072] (4-2) Obtain the baseline failure rate distribution function λ of each equipment in the workshop k (t)= , select the equipment M that affects the processing k A set of factors X = [X1, X2, …, X f , …, X F], and determine the influence of these factors on the equipment failure rate through design experiments β = [β1, β2, …, β F ];
[0073] (4-3) According to the equipment failure rate λ under time-varying working conditions rk (t|X) =αλ k (t)f(βX), represents the maintenance cost rate and equipment availability = Indicators, build equipment maintenance planning model V rk = By solving the derivative equation of the model and making the derivative equation equal to 0, the optimal maintenance time interval for each cycle of the equipment can be obtained. ; Among them, the preventive cycle index of device k is r, represents the duration of a preventive maintenance activity for equipment k, represents the unit cost of preventive maintenance activities, Indicates the duration of corrective maintenance activities, It represents the unit cost of corrective maintenance activities;
[0074] (4-4) According to the optimal maintenance time interval of the equipment, determine the time period for equipment k to perform maintenance activities in the rth cycle [ , ], determine whether the period is in the standby period of device k, if the actual start time of device k in the rth cycle maintenance activity is = Otherwise, further construct the cost balance of preventive maintenance activities and the cost balance of corrective maintenance activities, and optimize the actual maintenance time of the equipment based on the cost balance selection method;
[0075] (4-5) Assuming that the preventive maintenance activity of equipment k in the rth cycle is scheduled to be implemented before the zth position starts processing, construct the start time of the maintenance activity t a , when satisfied When t a Expressed as , when not satisfied When t a Expressed as ; The cost balance of corrective maintenance activities is expressed as , expressed as , which is equivalent to ; The cost balance of preventive maintenance activities is expressed as , expressed as ;in, represents the optimal maintenance time interval of equipment k in the rth maintenance cycle, represents the actual maintenance time interval of equipment k in the rth maintenance cycle;
[0076] (4-6) Assume that the preventive maintenance work of equipment k in the rth cycle is scheduled to be carried out after the completion of the zth position, and construct the start time of the maintenance activity t d , expressed as ; The cost balance of corrective maintenance activities is expressed as , expressed as , which is equivalent to ; The cost balance of preventive maintenance activities is expressed as , expressed as ;
[0077] (4-7) Based on the cost surplus optimization method and the economic benefits of advancing or delaying the equipment maintenance time, determine the actual maintenance activity time for each equipment. , the decision variable y rk(z-1) =1, =t a Otherwise, y rkz =1, =t d .
[0078] (4-8) With the goal of minimizing the total delay penalty cost, equipment load balancing penalty cost, and total maintenance cost, the initial production plan is optimized based on the aforementioned variable neighborhood search method, and the workshop equipment maintenance plan is simultaneously formulated, where the total delay penalty cost and device load balancing penalty costs The same expression as in the first-stage production scheduling model; the total maintenance cost is expressed as , TMC k is the maintenance cost of equipment k; the optimization objective of the second stage and the total maintenance cost are expressed as: ,in, TMC k = ∑ r ∊ B ∑ z ∊ Z y rk [ C k p + C k f × ∫ 0 T rk ' λ rk ( t | X ) dt ] ; ∀ k ∊ M ; where B={r1, r2, …, r K} represents the equipment maintenance cycle set, and the preventive cycle index of equipment k is r; u kz represents the number of failures that occur before the zth processing position of equipment k; W is a very large positive integer;
[0079] (4-9) Establish constraints to ensure that each device M k At most one maintenance activity can be performed on the system, expressed as: ;
[0080] (4-10) Establish constraints to ensure that the maintenance cycle of each device is arranged in sequence, expressed as ;
[0081] (4-11) Establish constraints to ensure that the actual start time of the equipment maintenance activity is later than the completion time of the current processing location process, expressed as ;
[0082] (4-12) Establish constraints to ensure that the start time of the next location of the equipment is later than the end time of the maintenance activity at the current location. , should meet the start processing time of the next position of the equipment Greater than or equal to the end time of the maintenance activity at the current location ;
[0083] The above method is applied to the production and maintenance plan formulation of the flexible operation workshop, where the maintenance-related parameters of each equipment are shown in Table 1, and the order task information and the processes that each equipment can process are shown in Table 2.
[0084] Table 1 Maintenance related parameters of each equipment
[0085] Table 2 Order task information and process information that each device can process
[0086] Firstly, the initial flexible job shop production scheduling scheme is generated based on the hybrid discrete particle swarm and variable neighborhood search method. The Gantt chart of the scheme is as follows: Figure 2 As shown, the total penalty cost for delay and the penalty cost for load balancing between devices is 1612.45 yuan;
[0087] In the initial production scheduling plan, the equipment failure rate is calculated under the two conditions of considering the working conditions and not considering the working conditions, such as Figure 3 As shown in Figure 2, the failure rate of each device increases with time. The failure rate curve of the device considering the working conditions presents dynamic and multi-level characteristics, such as Figure 3 As shown by the red line in the figure, the flexible workshop working conditions accelerate the probability of equipment failure;
[0088] Based on the maintenance decision model, the maintenance activity execution time of the workshop equipment under the initial production scheduling plan is analyzed, and the optimal maintenance activity time and the actual maintenance activity time are compared, such as Figure 4 As shown;
[0089] In this embodiment, the devices M8, M 10 and M 12 The actual maintenance activity execution period will still interrupt the normal processing of subsequent processes on these equipment. Next, the initial production plan is optimized based on the variable neighborhood search method, and the equipment failure rate and maintenance plan are updated simultaneously, so as to obtain a joint optimization plan for flexible job shop scheduling and maintenance decision-making with a total cost of 8665.865 yuan. The Gantt chart is as follows Figure 5 As shown;
[0090] In order to verify the effectiveness of the joint optimization method of flexible job shop scheduling and maintenance decision considering working conditions proposed in this invention, the scheduling strategies of other traditional flexible flow shops are applied to this case, and the total cost is compared. Strategy IV is the method proposed in this invention, and the other scheduling strategies are as follows:
[0091] Strategy I: Perform static regular maintenance on the equipment, and set the maintenance cycle T = 90 minutes. When the equipment operation time reaches T, directly insert the maintenance activity and mark the maintenance activity occupation period as unavailable.
[0092] Strategy II: Schedule maintenance activities based on the optimal maintenance time interval of the equipment. When the production task and maintenance activity conflict, execute the right shift strategy. The equipment will prioritize completing the processing of the current location process and postpone the maintenance activity.
[0093] Strategy III: Schedule maintenance activities based on the optimal maintenance time interval of the equipment. When production tasks and maintenance activities conflict, execute the left shift strategy, perform maintenance activities on the equipment first, and postpone the processing of the current location process.
[0094] The total cost of the production and maintenance solutions obtained by the above strategies is calculated, and the results are shown in Table 3. From the comparison of the total costs, it can be concluded that the joint optimization method of flexible job shop scheduling and maintenance decision-making considering working conditions proposed in the present invention can achieve significant total cost optimization.
[0095] Table 3 Comparison of total costs of production and maintenance solutions under each strategy
[0096] Compared with the existing technology, the performance indicators of this method are improved in that: a joint optimization method for scheduling and maintenance decisions of flexible job shops considering working conditions is proposed, specifically: an equipment reliability model considering time-varying working conditions is constructed, and an equipment maintenance planning model based on maintenance cost rate and availability index is proposed to solve the optimal maintenance time interval of each equipment. In view of the non-parallelism of production tasks and maintenance activities of the same equipment, a maintenance decision model based on cost surplus selection is proposed; for the first time, an integer programming model is established with the minimization of total cost including total delay penalty cost, load balancing penalty cost between equipment and total maintenance cost as the optimization goal, and the processing capacity of equipment in the flexible job shop, order task requirements, product process route, and production-maintenance coupling constraints are systematically considered; a two-stage joint optimization framework for scheduling and maintenance decisions of flexible job shops is innovatively constructed, and a highly economical and operational production and maintenance plan is obtained by solving the hybrid discrete particle swarm and variable neighborhood search method. Compared with traditional methods, the flexible workshop production and maintenance plan obtained by the present invention can realize real-time updating of equipment maintenance planning as the production plan changes, and can simultaneously optimize and adjust the production and maintenance plans according to the strong coupling relationship between production tasks and maintenance activities, effectively reducing the total delayed delivery, equipment load balancing penalty costs and maintenance costs.
[0097] The above-mentioned specific implementation can be partially adjusted in different ways by those skilled in the art without departing from the principle and purpose of the present invention. The protection scope of the present invention shall be based on the claims and shall not be limited by the above-mentioned specific implementation. Each implementation scheme within its scope shall be subject to the constraints of the present invention.
Claims
1. A joint optimization method for flexible job shop scheduling and maintenance decision-making considering working conditions, characterized in that: In the first stage, a production scheduling model is constructed according to the production capacity of workshop equipment, product process route and order task requirements, and the initial production plan is generated based on the hybrid discrete particle swarm and variable neighborhood search method; In the second stage, based on the maintenance decision model, the equipment maintenance needs under the production plan are analyzed and a maintenance plan is formulated. When the production tasks and maintenance activities on the same equipment conflict, the production plan is dynamically optimized based on the variable neighborhood search method and the equipment status is updated in real time. At the same time, the economically optimal production plan and maintenance plan are formulated.
2. According to the method of claim 1, the method is characterized in that: The production scheduling model is to establish an integer programming model with the goal of minimizing the penalty cost of order delay and the penalty cost of load balancing between equipment to obtain the optimal initial production scheduling plan, which specifically includes: Step 1: Define the parameters as follows: M = {M1, M2, …, M K } is the workshop equipment set, index k; P = {P1, P2,…, P I } is a product collection, indexed by i; i = {O i1 , O i2 , …, } is product P i The process set of M is indexed as j; ij For process O i Optional equipment set, process O ij Processing time PT ijk Depends on the selected device M K ; Z = {z1, z2, …, z K } is the number of processing positions of the equipment, indexed by z; D i For product P i Delivery period, For product P i Unit time penalty for delayed delivery; C b Penalty for load balancing between devices; Step 2: Define the decision variables as follows: ijkz ∈{0,1},x ijkz =1 indicates process O ij Processing at the zth position of equipment k, otherwise = 0; >0 indicates process O ij Start processing time; >0 indicates process O ij completion time; Step 3: Establish a model optimization goal to minimize the total cost. The total cost is composed of the total delay penalty cost and the equipment load balancing penalty cost. The total delay penalty cost refers to the sum of the delayed delivery penalties of all products in the order task. Indicates that TTC i is the delay penalty of product i. If the delivery time of the product is later than the required delivery time, a corresponding penalty cost is generated, which is proportional to the delay time. Otherwise, the delay penalty is 0. The device load balancing penalty cost is proportional to the load balancing level of all devices in the system, expressed as TBC. The smaller the variance of all device loads, the smaller the cost item. The optimization objective and the total cost of each item are expressed as: ,in , , ; Step 4: Establish constraints to ensure that each process of the product can only be assigned to one position on one piece of equipment for processing, expressed as: ; Step 5: Establish constraints to ensure that the processing of product processes strictly follows the process flow, that is, the start time of a process must not be earlier than the completion time of the previous process of the product, expressed as ; Step 6: Establish constraints to ensure that once a process starts, it cannot be interrupted before completion, expressed as ; Step 7: Establish constraints to ensure that device M k Each processing position can only be used to process one product at most, expressed as ; Step 8: Establish constraints to ensure that the processing position of each device is assigned to each process of the product in sequence, expressed as ; Step 9: Establish constraints to ensure that there is no overlap between processes assigned to the same equipment, that is, the start time of the current processing position of each equipment must not be earlier than the completion time of the previous processing position, expressed as ; Step 10: Based on the above parameters, decision variables, optimization objectives and constraints, the flexible job shop scheduling model is established as follows: ,in , ,in ; The first constraint of the model is , the second constraint is ,in , the third constraint is , the fourth constraint is , the fifth constraint is .
3. The method for joint optimization of flexible job shop scheduling and maintenance decision-making considering working conditions according to claim 1 is characterized in that: The hybrid discrete particle swarm and variable neighborhood search method comprises the following steps: Step a, initialization parameters, including particle swarm size, maximum number of iterations, crossover probability and mutation probability; Step b, a two-layer coding method is used to represent a particle or a feasible solution, wherein the two layers of coding are process sequence and equipment sequence, which are denoted as OS and MS respectively; the elements in OS represent the index of the processed product, and the number of times the index appears represents the number of processes of the corresponding product; the elements in MS sequence represent the processing equipment selected for each process in OS; Step c: Optimize the randomly generated initial particles by discrete particle swarm method, and update the particles with the objective function as the fitness value. The process includes three parts. The first part randomly selects the OS sequence or MS sequence of the particle for mutation operation. For the OS sequence, randomly select an element at one position and insert it into other positions to change the order of the OS sequence elements. For the MS sequence, randomly select multiple positions to change the processing equipment selection of the corresponding process, and generate new particles represented as ; Step d, the second part randomly selects an individual optimal particle in the crossover learning iteration process The first crossover operation is: randomly divide the order task P into two non-empty subsets P 1 and P 2 , from the current particle Take out the 1 The products in the collection are copied to the new particles, maintaining the element information of the process sequence and equipment sequence; Remove the 1 The product information in the original order is copied to the new particles. ; The second crossover operation is: randomly select the first S positions of the particle encoding sequence, and Take out the products in the first S positions in turn and copy them to the new particles, keeping the element information of the process sequence and equipment sequence; Remove the particles contained in the current The product information in the first S positions, copy the remaining products to the new particles in the original order ; Step e, the third part randomly selects a global optimal particle in the crossover learning iteration process The first crossover operation is: randomly divide the order task P into two non-empty subsets P 1 and P 2 , from the current particle Take out the 1 The products in the collection are copied to the new particles, keeping the element information of the process sequence and equipment sequence; in the global optimal particle Remove the 1 The product information in the original order is copied to the new particles. ; The second crossover operation is: randomly select the first S positions of the particle encoding sequence, and Take out the products in the first S positions in turn and copy them to the new particles, keeping the element information of the process sequence and equipment sequence; in the global optimal particle Remove the particles contained in the current The product information in the first S positions, copy the remaining products to the new particles in the original order ; Step f, when the iteration termination condition is reached, the particles with the top 10% of fitness values are selected as the input of the variable neighborhood search method; Step g: In the variable neighborhood search method, a deep local search is performed on the current particle through mutation and crossover operations. For the mutation operation, an element in the MS sequence is randomly selected to change the processing equipment selection of the corresponding process; for the crossover operation, a position in the OS sequence is randomly selected to insert the element at the position into other positions; Step h: When the iteration termination condition is reached, the particle with the best fitness value is output.
4. The method for joint optimization of flexible job shop scheduling and maintenance decision-making considering working conditions according to claim 1 is characterized in that: The maintenance decision model refers to considering the evolution of equipment failure rate under time-varying working conditions, combining the maintenance parameters of each equipment, and building a multi-objective optimization equipment maintenance decision model. The model is solved to obtain the optimal maintenance time interval of each maintenance cycle of the equipment, specifically including: Step ①: Define the parameters as follows: B={r1, r2, …, r K } represents the set of equipment maintenance cycles, the preventive cycle index of equipment k is r; the duration of a preventive maintenance activity of equipment k is expressed as , the corresponding preventive maintenance activity cost is ; When equipment k fails, the duration of corrective maintenance activities is expressed as , the corresponding corrective maintenance activity cost is ; The number of failures before the zth processing position of equipment k is recorded as u kz ; W is a very large positive integer; Step ②: Define the decision variables as follows: y rkz ∈{0,1},y rkz =1 means that the preventive maintenance activity is performed after the zth processing position in the rth maintenance cycle of equipment k, otherwise y rkz =0; >0 indicates the actual preventive maintenance time of equipment k in the rth maintenance cycle; Step 3: Define the intermediate variables as follows: rk (t|X) represents the failure rate function of equipment k in the rth maintenance cycle; represents the optimal maintenance time interval of equipment k in the rth maintenance cycle, represents the actual maintenance time interval of device k in the rth maintenance cycle, represents the optimal maintenance operation time point of equipment k in the rth maintenance cycle, represents the optimal maintenance end time of equipment k in the rth maintenance cycle, Indicates the actual maintenance operation end time of equipment k in the rth maintenance cycle; Step ④: Establish equipment reliability model under time-varying working conditions, λ rk (t|X) =αλ k (t)f(βX), where X=[X1, X2, …, X f , …, X F ] indicates the impact on equipment M during processing k A set of factors for the failure rate, β = [β1, β2,…, β F ] represents the coefficient of the corresponding factor in X; Step ⑤: Equipment M k The baseline failure rate function is represented by Weibull distribution λ k (t)= , where θ k and η k They represent the shape parameter and scale parameter in the Weibull distribution of device k, respectively; Step ⑥: Build equipment maintenance planning model V with the goal of minimizing maintenance cost rate and maximizing equipment availability rk = By solving the derivative equation of the model and making the derivative equation equal to 0, the optimal maintenance time interval of the equipment cycle can be obtained. ; Among them, the maintenance cost rate is defined as , equipment availability is defined as = , by respectively making d / dT rk = 0 and d / dT rk =0, the minimum maintenance cost rate value can be obtained and the maximum equipment availability value ; Step 7: Determine the time period for equipment k to perform maintenance activities in the rth cycle according to the optimal maintenance time interval of the equipment. , ], when the period is in the standby period of device k, the actual start time of the maintenance activity of device k in the rth cycle is = On the contrary, when [ , ] When the normal processing process of the process at the zth position of equipment k conflicts, the cost balance of preventive maintenance activities and the cost balance of corrective maintenance activities are constructed, and the actual maintenance time of the equipment is optimized based on the cost balance selection method; Step ⑧: When the preventive maintenance activity of equipment k in the rth cycle is scheduled to be implemented before the zth position starts processing, the start time of the maintenance activity is denoted as t a , when satisfied When t a Expressed as , when not satisfied When t a Expressed as ; The cost balance of corrective maintenance activities is expressed as , expressed as , which is equivalent to ; The cost balance of preventive maintenance activities is expressed as , expressed as ; Step ⑨: When the preventive maintenance work of equipment k in the rth cycle is scheduled to be carried out after the completion of the zth position, the start time of the maintenance activity is denoted as t d , expressed as ; The cost balance of corrective maintenance activities is expressed as , expressed as , which is equivalent to ; The cost balance of preventive maintenance activities is expressed as , expressed as ; Step ⑩: The cost surplus optimization method described above combines the economic benefits of advancing or delaying the equipment maintenance time to determine the actual maintenance activity time of each equipment. , the decision variable y rk(z-1) =1, =t a On the contrary, y rkz =1, =t d .
5. The method for joint optimization of flexible job shop scheduling and maintenance decision-making considering working conditions according to claim 1 is characterized in that: The production scheduling and maintenance decision joint optimization model is to minimize the total delay penalty cost, equipment load balancing penalty cost, and total maintenance cost, where the total delay penalty cost and device load balancing penalty costs The same expression as in the first-stage production scheduling model; the total maintenance cost is expressed as , TMC k is the maintenance cost of equipment k; the optimization target and the total maintenance cost are expressed as: ,in ; Specifically include: Step 1: Establish constraints to ensure that each device M k At most one maintenance activity can be performed on the system, expressed as: ; Step 2: Establish constraints to ensure that the maintenance cycle of each device is arranged in sequence, expressed as ; Step 3: Establish constraints to ensure that the actual start time of the equipment maintenance activity is later than the completion time of the current processing location process, expressed as ; Step 4: Establish constraints to ensure that the start time of the next position of the equipment is later than the end time of the maintenance activity at the current position. , should meet the start processing time of the next position of the equipment Greater than or equal to the end time of the maintenance activity at the current location ; Step 5: Based on the above parameters, decision variables, optimization objectives, constraints and the constraints in the first stage production scheduling model, the second stage production scheduling and maintenance decision joint optimization model is established as follows: The overall goal is expressed as ,in ,in ; The first constraint is , the second constraint is , the third constraint is , the fourth constraint is .in , .
Citation Information
Patent Citations
Equipment optimistic maintenance and production scheduling integrated optimization method under time-varying working condition
CN110865617A
Emergency order-oriented resource scheduling collaborative optimization method for flexible job shop
CN114021901A
Production scheduling and machine maintenance optimization method based on joint optimization model
CN114510874A
Scheduling and maintenance joint decision-making method for uncertainty multi-stage production system
CN115017811A
Preventive maintenance and production scheduling joint decision-making method for hybrid flow shop
CN115438815A
Cited By
Multi-agent-based dynamic flexible job shop multi-task scheduling method and system
CN121526249A
