一种参数自适应调整和边界约束的粒子群算法

粒子群优化算法的核心思想是每个粒子根据自己和周围粒子的"信息共享"寻优,达到全空间搜索最优解的目的。收敛速度快,全局寻优能力强。针对基本粒子群算法寻优精度较低,结果易发散的缺点,提出了一种参数自适应调整和边界条件约束的粒子群算法,惯性权重,学习因子随着迭代过程线性递加或递减,从而在算法初期个体能搜索整个空间,后期能够朝着全局最优值收敛而找到全局最优值。同时设置粒子边界条件约束,保证算法寻优解的准确性。理论分析和数值仿真结果表明了所设计方法的高效性,在保证算法效率的前提下,有效地提高了算法的寻优精度。

Particle swarm optimization with adaptive parameters and boundary constraints

The core idea of PSO is that each particle searches the best solution of optimization equations according to ＂information sharing＂ of surrounding particles and itself. PSO has fast convergence and high global search capability. For low accuracy and divergent results of fundamental PSO, this paper provides a kind of PSO with adaptive parameters and boundary constraints. Inertia weight and learning factors increase or decrease linearly with iterative process, in order to search the global space in early period of the algorithm and converge its global optimum in the latter. At the same time, the author set particle boundary constraints to ensure the optimization accuracy. Theoretical analysis and numerical simulation results show the efficiency and high optimization accuracy of the designed method.