Particle Swarm Optimization for natural grouping in context of group technology application

Abstract

Cell-formation problem (CFP) addresses the issue of creation of part families based on similarity in processing requirements and the grouping of machines into groups based on their ability to process those specific part families. The CFP is combinatorial in nature and due to difficulty faced in solving related mathematical programming problems; efforts have been made to use evolutionary approaches. Literature highlights that there are many advantages of converting batch type manufacturing system (BTMS) to cellular manufacturing system (CMS). In this paper, mathematical model has been proposed for groups to be emerged naturally. As mathematical model of CFP becomes NP- complete in nature, researchers advocate the use of meta-heuristics. Over the years, many different metaheuristic methods have been used to solve the CFP in group technology application. In the present paper, evolutionary population based method known as Particle Swarm Optimization (PSO) hybridized with assignment algorithm is used to solve cell formation problems. Due to these proposed changes, efficiencies of cell formed significantly increase in comparison to the results available in the literature. Proposed hybrid algorithm is applied to solve 30 different types of randomly generated and 10 standard CFPs, a large verity in terms of number of parts and number of machines required by these parts. For this algorithm, optimal values of parameters were also found with the use of Taguchi method. It is found that the proposed changes in algorithm and parameters obtained significantly impact the results in terms of efficiency values.