Simulated annealing and artificial immune system algorithms for cell formation with part family clustering

This study proposed a new model for cellular manufacturing systems to group parts and machines in dedicated cells using a part-machine incidence matrix to minimize the voids. After identifying the exceptional elements, the machines required for processing the remained operations of corresponding parts which are not processed in the dedicated cells are specified. This results in a new matrix called part family-machine. Then, by clustering the part family-machine incidence matrix, the part families that should be assigned to a specific cell to achieve the highest similarity can be determined. The similarity can be translated to sharing machines required for completing the processes and form new cells called shared cells to minimize the number of exceptional elements and voids. Unlike previous models in which the similarity is considered only in the dedicated cells, here, the similarity would be observed in the entire production process. Due to the model NP-hardness, two meta-heuristics including artificial immune system (AIS) and simulated annealing (SA) are proposed. The efficiency of the algorithms is compared to that of exact solutions. Also, the algorithms are compared regarding the quality of solutions. Finally, according to grouping efficacy measure, SA has a superior performance in comparison with AIS by spending less CPU time.