论文部分内容阅读
为了探索如何在协同进化算法中结合问题域固有的结构信息,实施全局稳定进展,本文首先分析问题维度所体现的个体间收益特征,提出一种维度识别方法.在此基础上,设计并实现一个协同进化算法.它能在进化过程中通过个体间的交互收益自动鉴别问题维度,并保存每个维度上当前已搜索到的最高测试值,以此作为评价基准控制进化在所有维度上均单调进展.配套设计的结构文档不仅有效支持维度鉴别,准确提供当前全局最高进展信息,而且存档量能达到最小化来保证算法的有效实施.模拟实验证实了该算法的可行性,并显示该算法较其它同类算法具有更高的性能和效率.
In order to explore how to combine the inherent structural information of the problem domain with the co-evolutionary algorithm and implement the global stability and stability, this paper first analyzes the characteristics of the individual income reflected by the problem dimension and proposes a method of dimension identification. Based on this, Co-evolutionary algorithm, which automatically identifies problem dimensions through the interaction between individuals during evolution, and saves the highest test values currently searched for in each dimension as a benchmark to control evolution in all dimensions. The structural design of supporting design not only supports the identification of dimension effectively and accurately provides the information of the current global maximum progress but also minimizes the amount of archiving to ensure the effective implementation of the algorithm.The simulation experiment proves the feasibility of the algorithm and shows that the algorithm is more efficient than others Similar algorithms have higher performance and efficiency.