一种改进的演化算法及其在求解复杂优化问题中的应用*

针对传统演化算法在求解函数优化,特别是多峰函数优化问题中出现的早熟现象以及演化后期收敛速度慢等问题,提出了一种新的反序小生境演化算法。该算法采用小生境反序交叉算子,以进一步增强局部寻优的能力;引入一种并行演化算法机制,加强群体寻优能力;同时,根据定义域划分初始种群,增加初始种群的覆盖面积。通过仿真实验表明,与传统的小生境演化算法相比较,利用该算法求解复杂多峰函数优化问题能够明显提高问题的求解精度和收敛速度,而且能够得到所有的全局最优解,更好地避免了求解问题时的早熟现象,达到了较好的效果。

Improved evolutionary algorithm and its application to solving complex optimization problems

Aiming at the problems of premature and slow convergent speed by using the traditional evolutionary algorithm in solving the optimization problems,especially in solving the optimization problems of multimodal functions,this paper proposed a niche inver-over evolutionary algorithm.This new algorithm adopted a niche inver-over operator to further enhance the local optimization ability.The algorithm introduced a mechanism of parallel evolutionary algorithm to strengthen the ability of swarm optimization.Besides,it divided the initial population heuristically to some sub-populations according to the domain for increasing the coverage area of initial population.Simulated experiments show that it enhances both solution precision and convergent speed more obviously by using this new algorithm to solve optimization problems of multimodal functions compared with the traditional niche evolutionary algorithm,and it avoids premature phenomenon better.