Study On Optimization Model And Algorithm For The Mixed-model U-shaped Disassembly Line Balancing And Seqencing Problem

Driven by the market of electronic information industry,the related technological innovation lead to the quicken upgrading of electronic products and many electronic products are scrapped when they are not in use,which results in a large number of E-waste.How to recycle and use it in a rational way has become a global problem.Thus,this paper studies the mixed-model U-shaped disassembly line balancing and sequencing problem(MUDLB/S)for multi-type electronic products.The content of this paper consists of the following two parts:For the cycle time(CT)as the main constraint of the MUDLB/S problem with stochastic task times,a mathematical model was established aiming at minimizing mean line idle rates,removing hazardous and high-demand parts as early as possible and minimizing mean the number of part removal direction changes.In addition,a hybrid multi-objective evolutionary algorithm based on decomposition and dynamic neighborhood search method(HMOEA/D)was designed to solve the problem to optimize the above objectives simultaneously.In HMOEA/D,a flexible tasks assignment strategy,dynamic neighborhood structure and dynamic weight vector adjustment were adopted to ensure the solutions' feasibility and the distribution of the non-dominated set.Finally,the algorithm was tested on benchmark instances generated by using Design of Experiment(DOE)techniques.Experimental results show that HMOEA/D is better than the parallel neighborhood search and the genetic algorithm based on local search.For the fixed number of workstations as the main constraint of the MUDLB/S problem with stochastic task times and positional constraint,a mathematic model was established aiming at minimizing cycle time(CT)and mean line idle rates.Besides,an improved parallel neighborhood search algorithm(IPNS)was proposed to optimize the above objectives lexicographically.In IPNS,two different kinds of neighborhood structure were defined,dynamic search strategy was adopted and paralleling was implemented by two different kinds of independent searches.Moreover,neighborhood exchanging directly was introduced to search the optimal solution.Finally,the simulation results show that IPNS is superior to PNS in terms of search efficiency and solution quality.