关键词: 锻造计划, 多阶段组批, 并行机, 启发式算法, 遗传算法, 重构

Abstract: Motivated by forge operation in a multi-layer production line whose jobs with uncertain process time in forge phase were processed simultaneously as a batch under certain condition, the objectives of minimizing the total tardy number of jobs and the total tardiness, two job-clustering-based algorithms were presented. In the first algorithm, the job priority was decided by process entropy of job, slacking importance of job and market will. And the batch priority was determined by the job with the highest priority in the batch. In the second algorithm, random keys were used to represent hybrid genetic algorithm, and random keys represent was adopted to perform coding and decoding. To solve practical big-size problems in a reasonable computational time, a rule coding method was adopted to optimize the gene sequence and crossover was also operated by anther rule. Niche technology was used to adjust the individual fitness, and elitist strategy was adopted to create next generation. Numerical experimental results showed that the hybrid genetic algorithm was better than the first algorithm, and the first algorithm was better than the other heuristic algorithms.

Key words: forging planning, multi-layer batch, parallel machine, heuristic algorithm, genetic algorithms, reconstruction