论文部分内容阅读
针对柔性作业车间调度问题,以总拖期最短为目标,提出了一种分层混合遗传算法。其中,根据总拖期的大小,将种群划分为精英层和普通层,精英层包含全局最优的数个不同质个体,其余个体划分为普通层;针对遗传算法局部搜索不足的问题,对精英层提出了一种邻域搜索策略,使代表机器选择和工序顺序的染色体可以根据自身的不足进行调节;针对遗传算法多样性容易丢失的问题,对精英层提出了一种灾变策略,不仅保留了种群的进化优势而且可以向优秀的个体学习。最后通过一系列标准测试函数以及一个生产中的实际案例验证了该算法的有效性。
Aimed at the problem of flexible job shop scheduling, aiming at the shortest total tardiness, a hierarchical hybrid genetic algorithm is proposed. Among them, the population is divided into the elite layer and the common layer according to the size of the total tardiness. The elite layer includes several individuals with the best global quality, and the other individuals are divided into common layers. For the problem of insufficient local search of the genetic algorithm, In this paper, a neighborhood search strategy is proposed, in which the chromosomes representing the machine selection and order of operations can be adjusted according to their own deficiencies. Aiming at the problem that the diversity of genetic algorithms is easy to be lost, a catastrophe strategy is proposed for the elite layer, The evolutionary advantage of the population can also be learned from outstanding individuals. Finally, the effectiveness of this algorithm is verified through a series of standard test functions and a production case.