具有自适应交叉算子的遗传算法及其应用

SHAPE OPTIMIZATION USING AN ADAPTIVE CROSSOVER OPERATOR GENETIC ALGORITHMS

简单遗传算法采用常数交叉概率和随机选择交叉点的方式进行交叉操作，这种操作方式带有一定盲目性和随机性，无法保证子代个体一定优于父代个体。为此提出了一个新的自适应交叉算子，依据每代个体的适应值函数来调整交叉位置和交叉概率，使杂交沿着有利于算法收敛的方向进行．为了验证这种自适应交叉算子的有效性和合理性，对一个二维多峰函数的极大值搜索问题，进行了求解．并将新算法进一步应用于离心叶轮的形状优化问题，结果表明具有自适应交叉算子的遗传算法在收敛速度和获得全局最优解的概率两方面都有很大提高。

Crossover operation is carried out using constant crossover probability and random interchange point in the standard genetic algorithm. This operation mode is blindfold and stochastic. It is not expected that the fitness value of sub-generation is larger than that of the parents. So an adaptive crossover operator is proposed, the location of crossover and cross probability is adjusted according to fitness function, so that cross operation is performed along with convergence direction. The improved genetic algorithm is applied to compute a two-dimensional multi-modal function and study shape optimization of a centrifugal impeller in order to verify algorithmic rationality and validity. It shows that convergence performance is greatly enhanced.