实数遗传算法进化策略的改进研究  

Improved Research on Evolutionary Strategy of Real Genetic Algorithm

作　　者：王吉权[1] 王福林[1] 董志贵[1] 田占伟[1] 文士发 

机构地区：[1]东北农业大学工程学院,黑龙江哈尔滨150030

出　　处：《数学的实践与认识》2017年第11期118-125,共8页Mathematics in Practice and Theory

基　　金：黑龙江省社科基金项目(16JYB06)

摘　　要：在现有文献研究的基础上,对传统实数遗传算法的进化策略又作了进一步研究,提出了一种改进的进化策略.进化策略克服了传统实数遗传算法中交叉得到的优秀个体有可能在变异过程中遭到破坏而不能生存的不足,并取消了交叉概率,使交叉产生的个体数增多,这样可增大产生更优秀个体的可能性,因而可使实数遗传算法的性能得到更好的改善.另外,给出了一种计算种群中个体适应度的计算公式和计算方法.该方法不但使得遗传算法具有较强的局部搜索能力,而且具有较强的广域搜索能力和较好的种群多样性,不易陷入局部最优解,从而可快速收敛到全局最优解.5个测试函数的计算结果表明,给出的实数遗传算法的改进进化策略比传统实数遗传算法进化策略的运算速度明显提高,迭代次数明显减少,从而验证了提出的实数遗传算法改进进化策略的有效性.Based on the existing research, the paper provides an improved evolutionarystrategy of real genetic algorithm, which makes further research on this field. The evo-lutionary strategy overcomes shortage of traditional real genetic algorithm whose excellentindividuals may not survive in the process of mutation, cancels crossover probability in orderto increase the number of individual produced in the process of crossover. This method canincrease possibility of producing excellent individuals, increases the performance of real ge-netic algorithm to get a better improvement. In addition, the paper gives calculation formulaand method on individual fitness of population. This method makes the genetic algorithmnot only have stronger local and wide area search ability, but also better population diver-sity. It＇s not easy to fall into local optimal solution that can make genetic algorithm quicklyconverge to the global optimal solution. The result shows that the improved evolution strat-egy presented in this paper has faster calculation speed and less number of iterations thantraditional real genetic algorithm by the trial calculation of five test functions, thus, theeffectiveness of improved evolution strategy is powerfully verified.

关 键 词：遗传算法 进化策略 交叉概率 适应度 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]