基于遗传算法的并行设计耦合集任务分布规划

Distribution Planning for Coupled Task Set in Concurrent Design Based on Genetic Algorithm

针对完全并行迭代模型在实际产品设计中难以实现的问题,从迭代过程中任务的迭代时间和迭代次数的有限性两个方面对迭代模型进行了修正,并在此基础上对迭代模型进行了扩展,建立了改进后的多阶段混合迭代模型。结合遗传算法,求解在多阶段混合迭代模式下耦合集的最短执行时间和最小成本以及相对应的任务分布方案。以某智能割草机的开发设计过程为例,说明了该方法的有效性,并分析了阶段数和学习率对任务执行时间和成本的影响。

Aiming at the problem of fully parallel iteration model which is difficult to implement in practical product design, this model was modified according to iteration time of task and the finitude of iterative number in the iterativc process. The improved multistage mixed iterative model was proposed on the basis of the extension of the modified model. Furthermore, Genetic algorithm was used to solve the shortest execution time and minimum cost as well as the corresponding task distribution scheme of the coupled task set for multistage mixed iteration model. Taking the design process of an intelligent lawn mower as an example, the effectiveness of this method was explained and the stage number of task influence on time and cost was analyzed.