Abstract: To combine the advantages of line stocking and kitting, a novel line-integrated supermarket model was proposed to further improve the part feeding process in mixed-model assembly lines, and the task assignment of logistic workers and part feeding period were optimized. First, the interdependent task assignment and cyclic part feeding problems were described. The mathematical models were built to minimize the total cost of logistic workers hiring and parts delivering. Then, based on the theorems, the dynamic programming embedded with heuristics was adopted to obtain the optimum for small scale problems. For medium and large-scale problems, a Modified Harmony Search Algorithm (MHSA) was constructed to generate satisfactory solutions. To accelerate the searching process, the harmony memory was repeatedly divided and regrouped. A local search method, crossover and mutation were employed to explore the searching space, thus to overcome the deficiencies of the original algorithm, such as limited search depth and the tendency to trap into local optimum. Finally, simulation of the modified algorithm was carried out and the results were compared with that of other evolution algorithms. The rapid operation speed and effectiveness of the proposed method were verified.

Key words: computer application, part feeding, line-integrated supermarkets, cyclic delivery, dynamic programming, harmony search algorithm