Application of fuzzy genetic algorithm for sequencing in mixed-model assembly line with processing time

Mixed-model assembly line (MMAL) is suitable for a production system that produces variable and changeable product models. Sequencing in MMAL is an important decision making since it has a direct impact to total assembly time as well as customer responsiveness. Various objectives are normally considered in MMAL. For instance, series-line flow-shop systems always employ minimizing the maximum flow time or minimizing makespan as their objectives. It is clear that in real world the processing time of each job is quite uncertain but conventional solution techniques normally omit this fact. To be more realistic, the processing time has to be formalized by using fuzzy set. This research proposes fuzzy Genetic Algorithms (fuzzy GAs) for sequencing in MMAL with fuzzy processing time. The objective is to maximize satisfaction of decision-maker, represented by fitness value of GAs. The sequence that maximizes the satisfaction is similar to that of minimizing the makespan. Furthermore, this study uses three different problems to compare the performance of fuzzy GAs with CDS (Campbell, Dudek and Smith) heuristic. Each problem differs in the number of products and minimum part set (MPS). Since the performance of fuzzy GAs depends on several parameters, pilot runs and experimental designs are used to test these parameters including population size, probability of crossover, probability of mutation, selection type, crossover type, and mutation type. Through performance comparisons, it is found that fuzzy GAs perform equally well or significantly better than the CDS heuristic. In addition, fuzzy GAs are a promising solution technique in searching for a good solution with an acceptable time limit.