聚类分析在多极值函数优化中的应用

将聚类分析方法应用于经典的优化和遗传寻优过程当中，提出了一种求解多极值函数全局最优解的方法。在基于梯度的算法中，先取多个初始点，几次迭代搜索后做聚类分析。在每类取一点，将目标函数分为多个单极值函数，然后分别寻优，通过比较得到全局最优解。在遗传算法中，通过聚类分析在每类取若干个体作为代表个体，它们将始终参与遗传操作，从而有望达到全局最优。

Application of cluster analysis in multi-extreme Valule function optimization

This text has applied the cluster analysis in the classic optimization and the genetic algorithm, and put forward a solution to find the global optimal value of the multi-extreme value function. In the algorithm based on the gradient, first, many initial points were selected; second, on the cluster analysis was made after several iterate searching; third, a value in each class is selected, thus the original function was changed into many single-extreme value function, and the optimal value was studied respectively in every function; and the last, the global optimal value was found by way of comparing the value. In genetic algorithm, select several individuals in every class were select with cluster analysis as the representatives, they will take part in the genetic operation all the time. In this way, we can find the global optimal value.