摘要
遗传算法是一种通过模拟自然进化过程搜索最优解的方法。但这种算法在求解最优解过程中总是以计算时间为代价来换得最优解的产生。对此,提出一种基于个体相似`性评价策略的改进遗传算法,融入了一种新的旋转交叉算子,每个子个体根据其与父个体的相似度和可信度来确定个体的适应度值,仅当可信度值低于某个阈值时,个体才做真实的适应度计算。实验结果显示,相似性评价策略计算得到的个体适应度值接近真实的适应度值,并且改进的算法求得最优解需要的评价次数明显要少于传统遗传算法,而在测试准测上的数据表明:提出的改进遗传算法相对于传统遗传算法,性能较好且求得的最优解也较为理想。
Genetic algorithm is a method of searching the optimal solution by simulating natural evolutionary process. But it always requires longer computation time for the best solution in solving process. This paper presents an improved genetic algorithm,it is based on individual similarity evaluation strategy. In it a new rotation crossover operator is incorporated. The fitness value of each individual is assigned according to its similarity and reliability with its parents. The real fitness of individual is only evaluated when the reliability value is below a threshold. Experimental results show that the fitness values of individual derived from similarity evaluation strategy are close to the actual ones,and the number of evaluations required for seeking the optimal solution by the improved genetic algorithm is significantly less than that of traditional genetic algorithm. Additionally,the data on test criterion show that the performance of the proposed algorithm and the optimal solution derived from it are relatively better than the traditional genetic algorithm as well.
出处
《计算机应用与软件》
CSCD
2016年第3期236-239,266,共5页
Computer Applications and Software
基金
国家自然科学基金项目(61202313)
江西省教育厅科研项目(GJJ13637
2013BAB211020)
关键词
遗传算法
相似性评价
交叉算子
Genetic algorithm
Similarity evaluation
Crossover operator
作者简介
汤可宗，副教授，主研领域：多目标优化。
张彤，本科。
罗立民，教授。
