遗传算法的改进及其在流体机械领域的应用  

Improvement on genetic algorithms and its application in fluid machinery

作　　者：陶海坤[1] 谭磊[2] 曹树良[2] 

机构地区：[1]武汉第二船舶设计研究所,湖北武汉430064 [2]清华大学热能工程系,水沙科学与水利水电工程国家重点实验室,北京100084

出　　处：《排灌机械工程学报》2010年第5期428-433,共6页Journal of Drainage and Irrigation Machinery Engineering

基　　金：国家自然科学基金资助项目(50776052);国家863计划项目(2008AA05Z201)

摘　　要：对遗传算法流程进行了多方面的改进,形成了可以称为'基于排序的轮盘赌、选优策略、启发式交叉、自适应非均匀性变异、跨世代精英选择、小生境策略'的遗传算法.该算法弥补了基本遗传算法容易产生早熟现象而得到局部最优解的局限性,加快收敛速度,计算选取的种群较小,降低了计算工作量.应用改进遗传算法,以大型泵站钟形进水流道为研究对象,结合CFD分析了设计参数与出口流速分布均匀度的关系.研究结果表明,改进遗传算法能成功地应用于钟形进水流道的多参数优化设计,改善其水力性能,2参数和5参数的目标函数值分别达到了96.817%和97.285%,具有很好的工程实际应用价值.GA（genetic algorithms） was overall improved： roulette wheel selection based on sorting,excellent selection strategy was used in selection operator,heuristic crossover was used in crossover operator,adaptive nonunformity variation was used in variation operation,also,cross-century elite choosing strategy and niche strategy were introduced.Traditional GA is easy to generate premature phenomena and get the local optimal solution.Improved-GA which overcomes the disadvantages of traditional GA speeds up the convergence rate.The population size selected by improved-GA is small,so as to reduce the calculation work.The relationship of design parameters and outlet velocity uniformity of bell-like inlet passage was researched using improved-GA and CFD.The results indicate that the improved-GA can successfully apply to multi-parameter optimization design of bell-like inlet passage.The hydraulic performance of bell-like inlet passage is greatly improved.Objective function values of two parameters and five parameters are 96.817% and 97.285%,respectively.This study provides references for engineering applications.

关 键 词：流体机械 遗传算法 优化 泵站 进水流道 

分 类 号：S277.7[农业科学—农业水土工程] TH311[农业科学—农业工程]