增加算子扰动项的粒子群优化算法研究  

Particle Swarm Optimization Algorithm Based on Increased Operator Perturbation

作　　者：刘文婧[1] 王建国[1] 吕达[2] 

机构地区：[1]内蒙古科技大学机械工程学院,内蒙古包头014010 [2]包头职业技术学院电气工程系,内蒙古包头014010

出　　处：《机械设计与制造》2017年第4期226-228,232,共4页Machinery Design & Manufacture

基　　金：国家自然科学基金项目(21366017);内蒙古科技大学科研启动项目(2014QDL024)

摘　　要：为了提高粒子群算法的稳定性,改善陷入局部最优的弊端,提出了一种增加算子扰动且对惯性权重进行正弦调整的粒子群优化算法。该方法首先利用差分方程对粒子的速度与位置变化过程进行深入分析,然后找到粒子群算法收敛的约束条件,进而获得改进后的惯性权重。最后在粒子群算法的速度公式中引入算子扰动项,其对粒子施加扰动,能够有效的抑制算法陷入局部最优问题,使算法在迭代后期也拥有一定的搜索能力。利用4个典型测试函数对算法进行验证,实验结果表明改进的惯性权重及速度更新公式使得该算法具备了较快的收敛速度和较佳的全局收敛性能,与标准粒子群算法相比,改进后的粒子群算法收敛精度高、鲁棒性强。To improve the stability of particle swarm optimization and to solve the problem of local optimum, a modified particle swarm optimization algorithm with increasing operator disturbance and the inertia weight with sinusoidal adjustment is proposed. Firstly, the changing processes of particle velocity artd particle position are analyzed by the method using difference equations and then the particle swarm optimization convergence conditions are got. Then the improved inertia weight is obtained. Finally, a disturbance term is added to particle velocity to prevent the algorithm into a local optimum solution and to make the algorithm in the late algorithm also have certain search ability. The improved algorithm performance is proven using four typical test functions. The experimental results show that the proposed method has faster convergence speed and better global convergence than the standard particle swarm algorithm And the improved algorithm has high convergence precision and strong robustness.

关 键 词：粒子群算法 算子扰动 惯性权重 正弦调整 差分方程 

分 类 号：TH16[机械工程—机械制造及自动化] TP181[自动化与计算机技术—控制理论与控制工程]