基于速度变异的粒子群算法

为了改善粒子群优化算法在收敛后期极易陷入局部最优的缺陷,提出了在非线性惯性权重策略粒子群算法的前提下,对陷入局部极值区域的粒子进行位置变异,使得粒子能很好地跳出局部极值区域,并在迭代前期及后期采用不同速度变异策略使处于个体极值点的粒子改变速度,能够有效地提高算法的前期全局搜索能力和后期局部开挖能力。通过4个经典测试函数验证了该算法具有更好的优化性能。

Particle swarm optimization based on velocity mutation

To improve particle swarm optimization getting into local optimum in the end of evolution stage,this paper proposes a strategy based on nonlinear inertia weight particle swarm optimization that in the iteration process the particles which fall into local optimum should carry out location variation,so that the particles can be well out of local extremum region.The particles in the locations of the personal optimum dynamically change their speeds with different strategies during the iteration process,which can effectively improve the global searching ability in the early stage and the capacity of local excavation in the late stage.Four classic benchmark functions demonstrated that this algorithm is of better optimum performance.