To study multi-objective flexible job-shop scheduling problem,the scheduling process in genetic algo-rithm(GA) is analyzed. A method for avoiding premature convergence is presented,an improved GA is designed. In population’ s individuals selection procedure,the elite individuals are first remained,then,non-dominated in-dividuals are preferentially selected,and the individuals which have less scheduling index value and larger crow-ding distance are preferential selected. Hence,complicated non-dominated sorting calculations are avoided. The contrast tests of algorithm show that the convergence performance of this algorithm is not lowered,and the algo-rithm can efficiently avoid premature convergence problem. It runs rapidly and steadily,and more different elite individuals which have the same scheduling index value can be obtained,then the scheduling worker’s selection range is enlarged.%为了研究多目标柔性作业车间调度问题,对遗传算法的调度过程进行了分析,提出了一种避免遗传算法早熟收敛的方法,并设计了一种改进遗传算法。种群个体选择过程先进行精英保留,再优选非支配个体,优选调度指标数值小且拥挤距离较大的个体;避免了复杂的非支配排序运算。算法对比测试表明,本文算法的收敛性能相当,能够有效避免早熟收敛问题;运行速度快而稳定;能够得到较多具有相同调度指标数值的不同精英个体,扩大了调度人员的选择范围。
展开▼