一种新的全局优化演化算法

A New Global Optimization Evolutionary Algorithm Based on Filled Function

演化算法在求解大型复杂多极值问题的过程中经常容易陷入局部最优,该文提出了一种变换目标函数法来消除早熟收敛。当演化算法检测出局部最优点时,使用填充函数构造变换目标函数,将局部极小点及其邻域提升,保留整体最小值点。从而新方法具有消除局部最优点而保留整体最优点的功能。通过对复杂的无约束优化问题和有约束优化问题的实验,结果显示了新方法具有搜索全局最优解的良好性能。

Finding the global optimum on a large,multimodal,complex landscape is usually very hard,even using the evolutionary approach.This paper proposes a new technique to alleviate the local minima problems.When evolutionary algorithm detects an undesired local minimum,the proposed method makes use of the filled function to escape.The filled function is defined as a transformed objective function to lift the neighborhood of a local minimum.It helps the global search method to eliminate local minima while preserving global ones.Experiments indicate that the new technique exhibits good performance and results in finding global minima.