基于单纯形的改进全局人工鱼群优化算法

Improved Global Artificial Fish Swarm Algorithm Based on Simplex Method

文中主要研究人工鱼群算法(AFSA)的优化问题。针对全局人工鱼群算法后期收敛速度慢、寻优精度低等缺点,在全局人工鱼群算法(GAFSA)的基础上,提出了一种改进的人工鱼群算法(MS_GAFSA)。该算法通过将全局人工鱼群算法与改进单纯形法相结合,以改善算法的收敛速度和寻优精度。MS_GAFSA首先以GAFSA进行迭代,利用GAFSA前期快速收敛及跳出局部最优值的优点收敛至全局最优点附近,此时以所在点为起点构造单纯形,并切换到改进单纯形法继续优化,通过反射、扩张、收缩和紧缩将单纯形翻滚、变形,快速收敛并趋近最优点,直至满足一定的精度条件停止,取此时单纯形上最优顶点值为目标函数最优值。通过对一系列benchmark测试函数的计算和比较,证明了该方法确实在寻优精度、收敛速度方面均有提升。

In order to overcome the drawbacks of Global Artificial Fish Swarm Algorithm （GAFSA） ,such as slow convergence and low precision optimization, a modified GAFSA （MS_GAFSA） is proposed, in which the modified simplex method is adopted to improve con- vergence precision and convergence rate. For GAFSA has a faster convergence in optimization of the early and the ability to recognize the local optimum value, a simplex is constructed based on the minimum given by GAFSA when the convergence turned to the stable point. Make the simplex move and roll by reflection, expansion and contraction. Compared the values of the simplex＇ s vertexes, constructing a new simplex by the trend of function, and repeating the process till the result is accurate enough. The computational results on benchmark functions show that MS_GAFSA achieves higher performance,including convergence precision and convergence rate.