Research On Adaptive Particle Swarm Optimization For Disassembly Line Balancing Problem

With the development of social economy,a large number of outdated and abandoned products have brought great pressure to the ecological environment.In the face of the problem of the waste of resources and environmental degradation that are caused by scrap products,people are determined to develop the sustainable industry and build a green ecological environment.Disassembly line can deal with a large number of waste products effectively and realize the recycling of scrap products on the strength of its high efficiency and informationization.Therefore,disassembly line is an important part in the whole green production process.However,there is disproportion of assigning the disassembly task to the workstation when the disassembly tasks become complex,which will increasing the production cost in the manufacture cycle.Based on the above analysis,the study of how to assign the disassembly task to the workstation,improve the disassembly efficiency and reduce the production cost is an important practical significance for the development of disassembly line.With the increasing of the problem scale that changes with the size of scrap products,the solution space will also increase,so it is difficult to obtain the optimal solutions for the traditional methods.By considering the characteristics of disassembly line balancing problem(DLBP),the particle swarm optimization(PSO)algorithm based on swarm intelligence is used to solve the complex DLBP,and the strategy that combines the search mechanism of particle with the operating mechanism of DLBP is used to ensure the feasibility of the optimization results in this thesis.The details of research contents are given as follows:1.The current research overview of the DLBP,the production factors and requirements used in the whole disassembly process are studied and the multi-objective disassembly line balancing problem model is established,which lay a theoretical foundation for the combination of improved PSO and DLBP.2.In order to improve the diversity of PSO,an adaptive dimension learning particle swarm optimization(ADL-PSO)algorithm is proposed for solving the DLBP.The method is based on enhancing information exchange among the particles.The conversion from continuous values to the disassembly task sequence is realized by constructing disassembly hierarchy information graph(DHIG),and DLBP is transformed into a problem of finding the best path according to the directed and weighted graph.In the ADL-PSO algorithm,an optimum dimensional individual is established and the dimension learning strategy is proposed to increase population diversity,which guarantee that ADL-PSO algorithm can get effective optimization solutions while solving complex DLBP.3.By introducing the concept of ecosystem,an adaptive cooperative particle swarm optimization(AC-PSO)algorithm is proposed to solve DLBP,in which the influence of external environment is considered and the parallel mechanism based on multi-population evolution is used.Considering the correlation between the disassembly tasks,the collaborative framework is improved by using the particle dimension stochastic decomposition strategy.In AC-PSO,the optimized vectors are randomly divided into several groups and each group is assigned a single population to optimize,then the greedy algorithm is used to select other parts as a whole to get the values of objective functions.The diversity of the AC-PSO is measured according to the population distribution entropy and the average particle distance,and the inertia weight is adjusted adaptively.The mutation operation is used to balance the exploration and exploitation ability of the AC-PSO algorithm.4.The performances of ADL-PSO and AC-PSO are tested on the nineteen standard DLBPs,SCH-3500 mobile phone and high-speed electronic tacking machine.The optimal results show that the ADL-PSO and AC-PSO algorithm are efficient to solve the complex DLBP.