Research On Production Scheduling Optimization Method Of Assembly Flow Shop Considering Preventive Maintenance

Due to its flexible and efficient production mode,assembly flow shop is widely used in manufacturing enterprises in the mass production of multiple varieties of products,such as fire trucks,computers,plastic products,clothing and automobiles.It mainly includes two stages:fabrication stage to process components and assembly stage to assemble these components into a qualified product.However,in workshop production,machines will inevitably be in some period of unavailability,such as maintenance operator and machine failure.Therefore,in order to ensure the continuity and efficiency of production and the high reliability of machines,this paper aims to integrate preventive maintenance(PM)and corrective maintenance(CM)into the assembly flow shop scheduling.Considering the diversity of the workshop layout,this paper gradually study the integration of two-stage assembly flow shop scheduling and PM,the integration of two-stage multi-machine assembly flow shop scheduling and PM,and the integration of assembly permutation flow shop scheduling and PM and CM.Finally,the validity and feasibility of the proposed model and algorithm are analyzed through practical cases.For the two-stage assembly flow shop scheduling considering PM,this paper aims to propose a new mixed integer linear programming(MILP)model to minimize the total completion time and maintenance time.meanwhile,12 constructive heuristics and 7metaheuristics are designed to solve the larger-scale instances.in the propose MILP model,each machine is given a new characteristic:maintenance level,and its initial value is determined according to the optimal maintenance interval of Weibull probability distribution.A new PM decision strategy is proposed to improve the current constructive heuristics,so that they can be used to determine the PM execution time points on each machine along with the product sequence.In the proposed meta-heuristics,three local search and two perturbations are proposed to improve the iterated local search(ILS)algorithm;Q-learning is first applied to determine the parameter combination in each iteration of ant colony algorithm,and select low-heuristic in the hyper-heuristic algorithm.By comparing the MILP model with the conventional model,it is observed that it is very important to integrate the flexible PM activities into the two-stage assembly flow shop scheduling.By comparing the 12 improved constructive heuristics,it was found that FAP?PM had the best performance in 720 benchmark instances.Finally,by comparing the7 meta-heuristics,it is observed that ILS1?Per2 outperforms the other 6 meta-heuristics.For the two-stage multi-machine assembly flow shop scheduling considering PM,the assembly machines were expanded from one to several sets.The maintenance level of each machine is determined in the same way as the previous problem.In order to ensure the reliability and production continuity of the machine,we need to find a suitable product sequence and PM execution time points,as well as the allocation of assembly machines.Therefore,this paper proposes a new MILP model,two constructive heuristics MCMT_PMand NEH_PM,and an PM-based iterated greedy algorithm.Finally,we designed three experiments to evaluate the performance of the proposed MILP model,heuristics,and meta-heuristic.The final experimental results show that the proposed MILP model can effectively solve the problems whose product numbers are less than 20;the performance of heuristic NEH_PM is much better than that of heuristic MCMT_PM,and proposed reference local search outperforms regular local search;Under the same termination condition,the proposed algorithm is superior to the other nine meta-heuristic algorithms.For the assembly permutation flow shop scheduling considering PM and CM,this paper proposes a new MILP model and Restarted Pareto Iterated Greedy Algorithm(RIPG)to minimize the minimum completion time and maintenance costs.Since the number of unexpected failures results in the nonlinear characteristics,the proposed model cannot be solved directly by the Cplex solver.Therefore,this paper proposes two lemmas to relax the expected failure numbers and CM cost,so that make the model linearized.In the proposed RIPG algorithm,a new solution evaluation method was designed to determine the maintenance plan and calculate the objective values.Four improvements,including bi-objective-oriented greedy and referenced local search phases,bi-objective-oriented acceptance criterion and restart mechanism,are designed to improve the performance of the algorithm.Finally,the MILP model and RIPG were evaluated through three types of experiments.The experimental results show that by using the Epsilon-constraint method,the MILP model is effective in solving small-scale examples;the proposed RIPG is superior to the other four multi-objective element heuristic algorithms.Finally,the proposed integrated optimization methods are used to solve the real engine production and assembly problems with DP_m?1|PM|,DP_m?P_m|PM|and DP_m?F_m|PM|forms.We first use the proposed MILP model to solve DP_m?1|PM|and DP_m?P_m|PM|problems,and then use the proposed RIPG algorithm to tackle DP_m?F_m|PM|form.By comparing the new scheduling and maintenance plans with the traditional plan,the results suggest that the proposed integrated optimization methods can effectively improve the production efficiency and reduce the maintenance costs in real engine production and assembly problems.Besides,,the management decision-maker can choose the appropriate plan to organize production according to the market situation and the actual work environments.