英语翻译遗传算法最早是由Holland J教授[4]提出的,它是一种模拟生物变异进化过程的全局收敛算法.但是传统的遗传算

英语翻译
遗传算法最早是由Holland J教授[4]提出的,它是一种模拟生物变异进化过程的全局收敛算法.但是传统的遗传算法在处理复杂优化问题或者存在多极值的函数优化过程中经常会收敛不到全局最优状态,容易陷入局部最优解.而对遗传算法性能影响最大的控制参数主要有：群体规模 、交叉概率 、变异概率 .其中 和 是决定遗传算法在搜索全局最优解时的性能的最关键因素.主要影响算法在搜索全局最优解时开辟新的解空间的能力,而 则主要影响算法搜索最优解区域并收敛到最优解的能力.因此,对不同的全局优化问题,和 都需要经过反复实验来确定,这一过程非常繁琐而且很难找到鲁棒性高的最佳选择.因此在本文的实际应用中我们选择了自适应的遗传算法作为求取最优特征选择的优化准则.该算法的优点是在遗传算法的操作过程中可以根据每一步的结果及时的调整交叉概率 和变异概率 .具体的定义我们在下面的章节里说明.

Genetic algorithm was first proposed by Professor HollandJ [4], it is globally convergent algorithm for simulating the evolution process of biological variation. But the traditional genetic algorithm in dealing with complex optimization problems or the presence of function optimization process of multipole values often can not find the global optimalconvergence, easy to fall into local optimal solution. Theperformance of genetic algorithm control parametersinclude: maximum population size, crossover probability,mutation probability. The genetic algorithm and is the most critical factor in the search for the global optimal solutionof performance. The main effect of algorithm in the ability of searching global optimal solution and open up newsolution space, which mainly affects the algorithm to search the optimal solution and converge to the optimalsolution ability. Therefore, the global optimization problem is different, and the need to determine through repeated experiments, the process is very cumbersome and difficult to find the best choice for high robustness. So in the practical application of this article we choose the adaptive genetic algorithm as the optimization criterion for optimalfeature selection. The advantage of this algorithm is totimely according to the results of each step to adjust the crossover probability and mutation probability in theoperation process of genetic algorithm. The specific definition of us in the following chapters illustrate.