混合变异粒子群优化算法及其应用

Hybrid Mutation Particle Swarm Optimization Algorithm and Its Application

为改善粒子群优化算法在解决复杂优化问题时收敛质量不高的不足,提出了一种改进的粒子群优化算法,即混合变异粒子群优化算法(HMPSO).HMPSO算法采用了带有随机因子的惯性权重取值更新策略,降低了标准粒子群优化算法中由于粒子飞行速度过大而错过最优解的概率,从而加速了算法的收敛速度.此外,通过混合变异进化环节的引入,缓解了粒子种群在进化过程中的多样性与收敛性这一矛盾,使得算法的全局探索与局部开发得到有效平衡.利用经典的基准测试函数和平面冗余机械臂逆运动学问题的求解来验证提出算法的有效性,试验结果表明:与其他算法相比,HMPSO算法具有更快的收敛速度、更高的收敛精度、更强的收敛稳定性以及更低的计算成本.

In order to improve the poor convergence quality of particle swarm optimization algorithm in solving complex optimization problems,an improved particle swarm optimization algorithm is proposed,namely the hybrid mutation particle swarm optimization algorithm(HMPSO).An inertia weight value updating strategy with a random factor is adopted in HMPSO,which reduces the probability of particles missing the optimal solution due to excessive flying speed,thus the convergence speed of the algorithm can be accelerated.In addition,through the introduction of hybrid mutation evolution,the contradiction between diversity and convergence of the swarm in the evolution process is alleviated,and the global exploration and the local exploitation of the algorithm are effectively balanced.The effectiveness of the proposed algorithm are verified by using the classical benchmark functions and the inverse kinematics of the planar redundant manipulator.The experimental results show that compared with other algorithms,HMPSO has faster convergence speed,higher convergence accuracy,stronger convergence stability and lower computational cost.