Research On Key Problems Of Assembly Sequence Planning Based On Double Sub-swarms Firefly Algorithm

Assembly sequence planning (ASP) is the core content of assemblyprocess design. An optimal assembly sequence has great significance on theimprovement of assembly efficiency, the reduction of assembly cost, theguarantee of assembly quality and shortens the development cycle. Mainresearch content of assembly sequence planning problem (ASPP) focus onassembly modeling, generation and optimization of assembly sequences andevaluation of assembly sequences.General approach to acquire the optimal assembly sequence is precisecalculation method and the intelligent algorithm based on the assemblymodel. Although the precise calculation method can ensure to acquire theglobal optimal solution, the combinatorial explosion is inevitable with theincrease in the number of components in products. When solving the ASPPbased on intelligent algorithm, it is difficult to cope with the strongconstraints and discreteness and sometimes the sequence we acquired maynot be practical. The ASPP was divided into three phases: acquired thefeasible sequences, sub-optimal sequence and the optimal sequence. And thena framework of ASP including assembly modeling, generation, optimizationand evaluation of assembly sequences was proposed based on doublesub-swarms firefly algorithm, interval inverse number method and TOPSIS.Firstly, a hybrid assembly model include the hierarchy structure,assembly relationships and parts information of products was constructedbased on the assembly graph model and assembly tree structure model. Theinformation required when generate, optimize and evaluate the assemblysequences was provided by the hybrid assembly model. Secondly, focus on the difficulties of strong constraints and discretenessin ASPP, a new discrete firefly algorithm (FA) based on double sub-swarmswas proposed. Double sub-swarms searching mechanism was presented tosolve the geometric constraints in ASP problem and a discrete FA wasintroduced to solve the discreteness problem in ASPP.Moreover, in order to make sure the assembly sequence is practical; ahierarchical evaluation index system composed of two types of indices(part-level and system-level) was presented. A method based on intervalinverse number was developed to evaluate part-level indices quantitatively.Meanwhile, assembly process simulation method for system-level indiceswas proposed. The candidate assembly sequences were ranked by TOPSIS toobtain the optimal assembly sequence.Finally, an example of aerospace product assembly sequences evaluationwas presented to illustrate the feasibility of the proposed approach.Acknowledgement：This dissertation is supported by Shanghai Scienceand Technology Innovation Action Plan Project (No.11DZ1120800).