Construction method of genetic algorithms for large-scale complex production scheduling problems

1.提出了一种能保持种群多样性的新选择方法。结果表明，所提出的选择方法（称为部分枚举选择法（Partial Enumeration Selection Method，简称PESM））能在不降低解的精度的情况下保持种群的高多样性。该算法对多个调度参数的变化具有较强的鲁棒性.分解和搜索空间缩减方法的提出对大规模调度问题的计算实验表明，所提出的分解和搜索空间缩减方法与遗传算法配合使用效果良好.一种能灵活处理复杂约束的遗传算法设计提出了一种遗传算法的设计方法，即只增加一个与所增加的约束有关的模块就可以解决调度问题。利用该算法求解了多台平行机作业车间的调度问题和作业车间的工人分配问题。针对无缓冲区作业车间问题，提出了一种新的译码方法.应用于多目标优化问题将基于PESM的遗传算法应用于求解多目标Flowshop问题，得到了光滑的Pareto字体。

1. Proposition of a new selection procedure capable of keeping a diverse populationIt was shown that the proposed selection procedure (called Partial Enumeration Selection Method : PESM) can keep a high diversity population through the generations without deteriorating the accuracy of the solution. The robustness of the algorithm was also shown for variations of several schedule parameters.2. Proposition of decomposition and search space reduction methodsIt was shown from computational experiments for large-scale scheduling problems that the proposed decomposition and search space reduction methods work well with genetic algorithms.3. Design of genetic algorithms capable of dealing flexibly with complex constraintsThe design of genetic algorithms was proposed in such a way that the scheduling problems can be solved by only adding a module related to the added constraints. The scheduling problems of a job shop process with parallel machines and the worker allocation problem in a job shop process were solved by the proposed algorithm. Moreover a new decoding method was proposed for no-buffer job shop problems.4. Application to multi-objective optimization problemsThe PESM-based genetic algorithm was applied to solving multi-objective flowshop problems and smooth Pareto fonts were obtained.