一种基于线性动态参数的自适应粒子群优化算法

An Adaptive Particle Swarm Optimization Algorithm Based on Linear Dynamic Parameter

提出一种权重因子和认知因子线性自适应性改变的粒子群优化算法(APSO-LDP),该算法中个体学习因子和社会学习因子都可以按设定的方式进行线性适应性改变。其中个体学习因子的线性减少、社会学习因子的线性增大,有助于粒子群前期的多样性和后期的跟随最优粒子,而惯性权重的线性减少更达到快速收敛和局部搜索能力的平衡。实验表明,该改进算法具有较好的寻优能力。

Presents an Adaptive Particle Swarm Optimization based on Linear Dynamic Parameter（APSO- LDP） algorithm. In this algorithm, the personal cognitive factor decreases linearly and the so- cial cognitive factor increases linearly, which can ensure the particle diversity in early search and follow the global best particle in late search. At the same time, the decreasing inertia weight achieves to keep balance between fast convergence and local search. Experimental result shows that the modified algorithm has better searching and convergence performance.