基于差异性激活方法的改进微粒群算法

Improved Particle Swarm Optimization Algorithms Based on Diversity-activation Approach

针对标准微粒群优化算法的惯性权重系数采用固定或线性递减的方式无法有效解决粒子陷入局部最优解的问题及可能出现的停滞现象,引入以差异性为基础的激活方法对微粒群算法进行改进.在每次迭代时算法可以动态调整惯性权重参数及粒子的活性,从而促进粒子收敛至全局最优解.对6种典型函数的实验结果表明,引入本文的激活方法后,改善了微粒群算法的开发和探索能力,并提高了其收敛速度及精度,其中以非线性惯性权值递减策略的微粒群算法最为明显.

In the classical particle swarm optimization algorithm, a constant or linearly decreasing inertia weight was used for solving the optimization problem, but it could not solve the phenomenon of stagnation. A diversity-based inertia weight strategy and the activation of swarm in the particle swarm optimization were proposed. In each iteration process, the inertia weight and activation of swarm were changed dynamically, which benefits to the algorithm converging quickly to global optimal solution. According to the experimental results using six typical functions, the activation approach for the particle swarm optimization improves exploitation and exploration ability, but still keeps a rapid convergence and fine precision, and the nonlinear strategy for decreasing inertia weight of the particle swarm optimization is the most obvious.