Abstract: For Just-in-Time (JIT) part distribution problem in Mixed-Model Assembly Lines (MMALs), an integrated vehicle filling and routing scheduling model is presented with consideration of loading capacities and no sock-outs constraints of assembly lines. First, a scheduling problem domain is described and then the mathematical programming model is developed with an objective to minimize the maximum weighted inventory level in all stations and production cycles. Second, the backtracking method is established to optimally solve small-scale instances based on two properties of the proposed scheduling model. Third, a Modified Discrete Artificial Bee Colony (MDABC) algorithm is introduced to cope with the exploding search space of medium- and large-scale instances. Both local search and differential evolution operations are employed for neighbor search, which improves the convergence performance of the modified metaheuristic effectively. Finally, simulations are carried out and the experiment results verify the feasibility and effectiveness of the proposed JIT distribution model and methods.

Key words: artificial intelligence, mixed-model assembly lines, backtracking, artificial bee colony