一种新的快速进化规划方法

An Improved Fast Evolutionary Programming

针对快速进化规划方法存在的问题——进化搜索过程中存在较多非法个体 ,该类个体变量位于问题定义空间之外 ,造成大量无效的计算开销。提出了在进化过程中及时剔除非法个体 ,只允许合法个体参与竞争选择的进化规划方法 ,实验及分析结果表明 ,在复杂函数优化问题的进化求解中 ,该方法能有效提高进化求解速度和效率。

Fast evolutionary programming (FEP) with Cauchy mutation, first proposed by Yao et al in 1996 [1], was discovered to be still needlessly slow by us. We found that some of the mutated individuals fell outside of the defined space and that these unwelcome mutated individuals could vary from none to as much as half of all the mutated individuals. We call such unwelcome individuals illegal individuals. In our algorithm, we eliminate all illegal individuals as soon as they are discovered in order to save unneceassary operations. Experimental results, as shown in Fig.2 and Table 1, show preliminarily that our IFEP (improved FEP) can increase the speed and efficiency of evolutionary computation significantly.