基于遗传算法和模型仿真的调度规则决策方法

Selection of scheduling rules based on genetic algorithm and process simulation

为了完成特定生产环境下的调度规则选择问题,提出一种将遗传算法和过程仿真相结合的调度规则求解方式。在该求解方式中,遗传算法采用分段整数编码,每个染色体都代表一组可用于描述具体调度方案的规则组合;遗传操作包括选择、交叉、变异三种类型;为获得适应度函数值,利用基于某扩展Petri网的生产过程模型进行仿真,以在每一代种群中,得到与每个染色体相对应的各项性能指标值,进而以一种集成层次分析法和方案模糊评判的决策优化方法求取相应的适应度函数值。另外,为了改善串行遗传算法不切实际的解答时间,用主从式并行遗传算法代替传统遗传算法,保证了解在时间上和质量上的可行性。

To obtain the best scheduling rule set in specific manufacture environment, a method integrating the genetic algorithm (GA) and process simulation was proposed. In this method, each chromosome representing a rule set was formed by combining several integer segments, and genetics operations involving choice, crossing and variation were performed and they act according to certain order. To get adaptive function value of every chromosome, some extend Petri net based on the production process model was built, all the performance index values of every chromosome were gained by model simulation and an analytic hierarchy process (AHP) method combining with fuzzy evaluation was used to calculate adaptive function value. Furthermore, for lessening calculation time of GA, the traditional serial GA was replaced by the parallel GA. The validity and practicability of the method were verified by an example.