【摘要】 基于混沌序列的多峰函数微粒群寻优算法的目标就是找到多峰函数的所有局部优化峰值。在分析微粒群优化算法中各个参数对微粒运动影响的基础上,对微粒群算法进行改造,让微粒运动从初始位置沿优化函数曲线向优化峰值方向爬行,直至找到所在区域的局部优化峰值;要想求得尽可能多的局部优化峰值,就要求微粒群中微粒的初始位置分布具有随机性和遍历性,为此采用混沌序列设置微粒初始位置;为使每一个局部最优值点都可能有微粒群中的微粒经过,采用变步长的迭代计算;为防止优化函数曲线的某些局部峰附近没有微粒分布,从而漏掉该局部峰值,对计算进行重复,直至两轮求得的优化函数的局部峰值之差小于给定阈值。仿真结果表明,该算法具有很好的局部寻优特性,计算过程简捷,寻优效果良好,可有效地应用于多峰函数的局部寻优并求取全局最优值。更多还原

【Abstract】 It makes a searching for all local optimization of the multimodal function that a PSO algorithm based on chaos sequence for multi-modal function optimization.On the foundation of analysing every parameter of PSO algorithms,PSO algorithm is reformed and particles move from initialization position along multimodal function super-surface toward local optimization direction until reach nearest local optimization.In order to find most local optimization,it is necessary that particle initialization position is random and spread over range of variable,thus chaos sequence is used in particle initialization.For the sake of the randomicity and ergodicity of chaos sequence,alternate compute is not until last twice calculate difference is smaller than set a value.A computer simulation result shows that the method is very well for local optimization.Not only the method is simpler expression and easier operation,but also can give the all local optimization and a global optimization value.The method can be conveniently and efficiently used in multi-modal function optimization.更多还原