Abstract: Concerning the problem that the standard Particle Swarm Optimization (PSO) algorithm in which inertia weight is global parameter, and cannot adapt to the complex and nonlinear optimization process, a Spacing Feedback Inertia Weight (SFIW) was proposed. Taking advantage of the characteristic that Sigmoid function can make the smooth transition between linear and nonlinear, an inertia weight function based on Logistic equation was constructed. In the process of optimization, the nonlinear coefficient of inertia weight function was adjusted according to the particle spacing to make the particle with longer particle spacing get larger inertia weight and make the particle with shorter particle spacing get smaller inertia weight. Therefore, the local exploitation and global exploration get balanced. Finally, the experimental results on several benchmark functions and the comparison with other algorithms show the effectiveness and feasibility of the SFIW-PSO.

Key words: Particle Swarm Optimization (PSO), inertia weight, particle spacing, Sigmoid function, local exploitation, global exploration

摘要： 针对标准粒子群优化（PSO）算法把惯性权值作为全局参数,很难适应复杂的非线性优化的问题,提出了一种基于粒距和S型函数的粒子群权值调整策略（SFIW）。利用S型函数能够在非线性和线性之间平滑过渡的特性,构造了基于Logistic方程的惯性权值函数。在优化过程中根据每个粒子的粒距大小,调整每个粒子的惯性权值函数的非线性系数,使得粒距较大的粒子获得较大的惯性权值、粒距较小的粒子获得较小的惯性权值,从而平衡算法的局部开发和全局探测能力。最后,通过对基准函数的仿真并与其他PSO算法比较,验证了算法的有效性和可行性。

关键词: 粒子群优化, 惯性权值, 粒距, S型函数, 局部开发, 全局探测