PARETO SIMULATED ANNEALING FOR BALANCING THE MULTI-OBJECTIVE ASSEMBLY LINE TYPE II PROBLEM WITH SEQUENCE-DEPENDENT SETUP TIMES BETWEEN TASKS

Line balancing is a fundamental concept for continuous production systems. Assembly lines are present in different industrial environments and usually have a great economic impact because of their high manpower levels. A simpli ed view of the assembly line balancing problem (ALBP) is de ned as the grouping of the tasks required to assemble the nal product to the workstations conforming to the assembly line, which speci es the permissible orderings of the tasks. The main goal of the assembly line balancing problem is to assign the tasks to workstations such that the precedence relations are satis ed and some performance measure is optimized. The ALBPs are classi ed into two groups: simple assembly line balancing problems (SALBPs), which bear numerous simplifying assumptions, and general assembly line balancing problems (GALBPs), which are closer to reality due to the consideration of one or more realistic conditions, like sequence-dependent setups.
In this paper, we consider the problem of optimizing simultaneously the objectives of minimizing cycle time and minimizing the overall setups in a general assembly line balancing environment with the consideration of sequence-dependent setup times between tasks. The rst objective, which is referred to as the type II problem, generally occurs when the organization wants to produce the optimum number of items using a xed number of workstations without adding new machines. The minimization of the overall setup times is important mostly for the cases when setups impose maintenance costs on tools and the prolongation of setup times would increase maintenance costs and also bring more exhaustion to workers.
This paper is intended to introduce the objective of minimizing overall setups in the class of assembly line balancing problems and solve the problem of concurrently minimizing cycle time and the overall setups. The exact method was not ecient enough to solve the innovative problem with type II problem assumptions; thus, a Pareto simulated annealing (PSA) algorithm is developed to solve such an Np-hard problem and several quantitative metrics are de ned for evaluating the proposed algorithm. Computational results veri ed the considerable eciency of the PSA algorithm.