A new multi-objective genetic algorithm for solving the fuzzy stochastic multi-manned assembly line balancing problem

By incorporating the uncertainty and imprecision inherent in real-world production systems, this paper introduces the multi-manned assembly line balancing problem (MMALBP) with fuzzy stochastic task processing times. MMALBP has recently emerged as a key challenge in flow-line production systems manufacturing large-scale products (e.g. the automotive industry), where multiple workers collaborate at the same station to perform different operations simultaneously on a single product. MMALBP is a decision problem that involves partitioning assembly tasks among stations and scheduling them across multiple workers while optimising key operational objectives related to capacity and/or operational cost of the line. To enhance realism and decision-making, a new problem termed fs-MMALBP is introduced, which models task time uncertainties as fuzzy stochastic variables. The objective is to optimise three conflicting criteria: (1) minimising the number of the stations, (2) minimising the total number of the workers employed along all the stations and (3) maximising the workload smoothness across the line. Given the NP-hard nature of the problem, a new robust multi-objective genetic algorithm (MOGA) is developed to identify the Pareto-optimal set. Simulation results show that MOGA effectively produces well-distributed Pareto-optimal solutions under fuzzy-stochastic uncertainty, with improved hypervolume and competitive CPU times across benchmark instances.