Research On Generalized Cell Formation Problem Considering Inherent Attributes Of Parts

With the increasing competition of market and customization-oriented demands,cellular manufacturing(CM)has become an effective technique for companies nowadays to manufacture various products in small or medium size batches with lower cost.Essentially,CM is a method of attaining great economic benefits through clustering parts and machines into groups,respectively.As one of the most important steps for constructing a cellular manufacturing system,the aim of cell formation is to cluster similar parts into families and relevant machines into groups.A generalized cell formation problem is proposed,considering alternative process routings and operation sequences of parts.The linearized model is easily solved by CPLEX(v12.5)to find exact solutions.To improve the computing efficiency,an automatic approach based on improved discrete Cuckoo Search(IDCS)algorithm is first designed to address the proposed cell formation problem.In IDCS,Lévy flights are redefined to perform the evolution of discrete solutions.Local search is embedded to improve the searching ability.The computational results of six testing problems are compared with those obtained by genetic algorithm-based and particle swarm optimization-based approaches,which reveals that IDCS is much more effective,efficient and robust.Most existing models and approaches are based on production flow analysis.Their applicability can be extended by taking inherent attributes of part into consideration,such as geometric and material features of parts,which directly influence fixture and tooling management.This paper investigates a cell formation problem integrating production-flow and inherent attributes of parts.Symbolic and fuzzy data are utilized to describe inherent attributes.A bi-objective model is proposed,in which the first objective is to minimize intercellular movements based on production flow analysis,and the second objective is to minimize dissimilarity between inherent attributes of parts within a family.Given the computational difficulty of the investigated problem,a novel approach based on hybrid discrete Cuckoo Search(HDCS)algorithm is first designed to address this problem with large scale.A discrete version of Lévy flights is defined to perform the evolutions in the solution space.Computational experiments are conducted and Pareto optimal sets are obtained to provide alternative solutions for decision-makers.By solving a set of small to large instances,multi-objective metrics of HDCS and NSGA-II are calculated.Comparison of the results shows the higher performance of HDCS.