面向汽车投产排序的混合多目标网格遗传算法

汽车投产排序时，希望同时实现零部件消耗均衡化、车型调整费用最小化、工位作业位置精准化三个目标，为此提出一种基于Pareto层级的混合多目标网格遗传算法(HmoGA)。先将个体排斥机制加入到Pareto层级构造中，使非支配解的分布更均匀,再融合Pareto层级划分、网格拥挤度评价与相邻个体几何距离计算，设计一种多目标自适应网格选择机制，用于从动态变化的父代种群中选择较优个体构成进化种群、获取交叉运算的父代基因、改善非支配解集的分布质量。混合双基因位的迁移算子对非支配解进行邻域搜索，适时扩大搜索空间，跳出局部最优。利用三组不同规模的测试问题集,从非支配率、非支配解数量和相邻个体距离偏差三个指标方面进行比较,实验证明HmoGA算法在收敛性、解的数量和分布性方面都比NSGA-Ⅱ算法有显著优势。

A Hybrid Multi-objective Grid Genetic Algorithm for Automobile Production Sequencing Problems

Three objectives were expected to be achieved simultaneously when sequencing automobiles in process,including equaling the spare parts consumption,minimizing the total  adjustment cost resulting from exchanging automobile models,calibrating the work position for each  automobile on any station.A new hybrid multi-objective grid genetic algorithm(HmoGA) was proposed based on  Pareto stratum. In the algorithm,a new rejection mechanism was first  considered in the sorting process of Pareto stratum,for the purpose of getting the even distribution of the non-dominated solutions.Then an adaptive grid selection scheme was  designed by integrating  Pareto stratum evaluation,crowding  degree  calculation and distance estimation among adjacent individuals.Thus,higher quality population can be generated,better parent chromosomes can be  selected,and the distribution of the Pareto front can be improved constantly. Finally,the 2-opt shift operator was hybridized into the proposed genetic algorithm so as to broaden the search space and escape from local optimum.Three groups of experiments have done and three metrics including  non-dominated ratio,the number of Pareto optimal solutions and the deviation of distances  between neighbors were used as the performance measures.The results reveal that the proposed HmoGA dramatically outperforms NSGA-Ⅱ in terms of convergence and diversification.