灰熵并行分析法多目标次序敏感性研究

针对多目标优化算法对目标次序表现敏感这一特点,提出对灰熵并行分析法的目标次序敏感性进行分析.以多目标流水车间调度问题为对象,建立不同目标次序的多目标流水车间调度模型,以基于灰熵并行分析的GA优化不同次序的多目标模型.定义敏感系数,绘制敏感系数曲线图.实验结果表明,灰熵并行分析法对目标次序敏感,按目标值大小排列的升序和降序并不是最好的次序,震荡次序敏感系数最小.表明多目标优化时要选择合理目标次序以获得更好的优化结果.算法适应度值同样对目标次序敏感,对算法搜索效果影响明显.

Multi-Objective Order Sensitivity Study on Grey and Entropy Parallel Analysis Method

In view of the characteristic that multi-objective optimization algorithms are sensitive to objective order, objective order sensitivity analysis based on grey and entropy parallel analysis is proposed. Taking multi-objective flow shop scheduling problem as object, flow shop scheduling model of different objective order is established, and GA based on grey and entropy parallel analysis is used to optimize the multi-objective model of different objective order. Then sensitive coefficient is defined and graphics of sensitive coefficient is drew. The experiment results show that grey and entropy parallel analysis is sensitive to objective order, ascending and descending order ordered by the size of objective value are not the best, and sensitive coefficient of shock order is the smallest. It indicates that we should select reasonable objective order to get better optimization results for multi-objective optimization. Also algorithm fitness value is sensitive to objective order, which effects significantly on the algorithm performance.