基于均匀设计的并行变异遗传算法

针对标准遗传算法易陷入局部最优和局部搜索能力差的缺陷,提出一种基于完全均匀设计的并行变异遗传算法(U D-PMGA)。首先用完全均匀设计方法获得多样性良好的初始种群,之后提出两种改进的锦标赛选择机制,一种改进方法取适应度较高的50%个体覆盖整个群体,另一种改进是用适应度较高且各自不同的50%个体覆盖整个群体。把适应度相近且空间距离较远的两个个体作为交叉对象以提高交叉操作的搜索效率。最后用自适应变异比例和自适应变异步长相结合的并行变异搜索策略提高算法的收敛速度和搜索精度。通过测试函数仿真,并与其它算法比较,验证了UD-PMGA算法的有效性与先进性。

Genetic Algorithm with Parallel Mutation Based on Uniform Design

Aiming at the defect that standard genetic algorithm is easy to fall into local optimum and poor local search ability,a genetic algorithm with parallel mutation based on fully uniform design(UD-PMGA)is proposed.First,obtaining the initial popula tion of good diversity with completely uniform design method is obtained,then two kinds of improved tournament selection mecha nism is proposed.One improved method is that the entire population is covered by 50%individuals with higher fitness,the other is that the entire population is covered by 50%different individuals with higher fitness.Two individuals with close fitness and far dis tance are used as cross objects to improve the search efficiency of cross operation.Finally,parallel mutation search strategy with the mutation of adaptive ratio and adaptive steps is proposed to improve the convergence speed and the search precision.Through the test function simulation,and compared with other algorithms,the validity and advance of the UD-PMGA algorithm are verified.