研究了单机环境下工件尺寸有差异的批调度问题,设计了一种改进蚁群算法对问题的制造跨度进行优化。首先引入了Metropolis准则的概率选择机制作为路径激励策略,避免蚁群算法过早收敛的问题;然后采用了Batch First Fit算法对蚁群的路径进行解码,以产生可行的分批方案.最后选取了问题的所有24类算例,将改进的蚁群算法和遗传算法及模拟退火算法进行了全面的对比实验,结果验证了改进的蚁群算法的有效性。
An improved Ant Colony Optimization (ACO) method was proposed to minimize the makespan on a single batch-processing machine with non-identical job sizes. The Metropolis criterion was applied to modify the selection mechanism of paths to avoid immature convergence of ACO. To decode the paths of ants, Batch First Fit heuristic was used to transform the paths into feasible batches. In the experiment, the improved ACO was compared to Genetic Algorithm and Simulated Annealing on all the 24 levels of instances. The results demonstrate the efficiency of the improved ACO.