摘要
非支配排序遗传算法用于气动发动机设计不能获得完整的功率与比功关系曲线,为此对程序中的等级排列子程序和分散性估计方法进行了改进.将两目标优化问题中的性能指标分别定义为空间性能指标和跟随性能指标.通过一个区间分布参数将空间性能指标分成多段,位于同一区段内的个体根据其跟随性能指标的大小进行等级排列.个体间的分散性只根据空间性能指标进行计算.通过对预先设计的以正弦函数为目标的优化问题进行求解,验证了改进后的程序能够获得准确、分布均匀的解.与NSGA-II算法相比,改进后的程序用于气动发动机设计可以得到更加完整的设计信息.
Multi-objective optimization program NSGA-II (non-dominated sorting genetic algorithm) cannot achieve the complete correlation between power and specific work in air powered engine designing. This work modified the ranking and density estimation subprograms of NSGA-II. In the two-objective optimization problem, objectives were grouped into space performance objectives and following performance objectives. A sharing parameter was designed to divide the space performance objective into small sections. Individuals in the same section were ranked according to their following performance. The crowding distance was calculated based on the individual's space performance. Application to a designed sinusoid function ob- jective optimization showed that the modified program can get accurate and evenly distributing solutions. Compared to the original NSGA-II, the modified optimization program can present more integrated information for the air powered engine design.
出处
《浙江大学学报(工学版)》
EI
CAS
CSCD
北大核心
2009年第5期907-910,共4页
Journal of Zhejiang University:Engineering Science
基金
国家教育部博士点专项基金资助项目(20020335079)
关键词
气动发动机
多目标优化
遗传进化算法
air powered engine
multi-objective optimization
genetic evolutionary algorithm