论文部分内容阅读
本文就遗传算法解决多极值函数优化问题的方法进行了分析总结，提出了一种基于多生境排挤机制的改进算法，并将其应用到ＭＵＳＩＣ谱的谱峰搜索过程中。实验证明，改进后的方法较原算法计算效率更高，能够在一次搜索过程中同时确定目标函数多个极值点的位置。对于ＭＵＳＩＣ谱谱峰搜索问题，计算量较等步长遍历搜索节省在两个数量级以上．
In this paper, the method of genetic algorithm to solve the multi-extremal function optimization problem is analyzed and summarized. An improved algorithm based on multi-habitat crowding mechanism is proposed and applied to the peak search of MUSIC spectrum. Experiments show that the improved method is more efficient than the original algorithm and can determine the location of multiple extremum points of the objective function simultaneously in a search process. For the MUSIC spectrum peak search problem, the computational cost is more than two orders of magnitude compared with the step size traversal search.