| 摘要: |
| 针对粒子群优化算法中出现的收敛早熟和不收敛的问题，提出了一种基于自然选择和惯性权值非线性递减的改进粒子群算法，在算法迭代过程中，粒子边界速度采用最大速度非线性递减变化策略来限制，惯性权值非线性递减变化用于平衡种群粒子前期全局搜索与后期局部寻优的能力；为使种群在进化过程中保持多样性，在标准粒子群算法中引用二阶振荡策略使种群在进化过程中始终保持着多样性；在此基础上，进一步地将遗传算法中的选择机理与粒子群算法结合起来用于提高算法的适用性能；所提出的算法经过多个基准测试函数的模拟实验验证，并与其他已有算法进行了对比；实验结果表明:算法在搜索精度与寻优能力上有更明显的优势，尤其是在多维、多峰等复杂非线性优化问题时，所提算法具有很强的竞争力。 |
| 关键词: 粒子群优化算法 惯性权值 自然选择 最大速度非线性递减 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| An Improved Particle Swarm Optimization Algorithm Based on Nonlinear Decreasing Inertial Weights |
|
HUA-Yong，WANG- Shuang-yuan，BAI Guo-zhen，LI Bing-chu
|
|
School of Mechanical Engineering，University of Shanghai for Science and Technology，Shanghai 200093，China
|
| Abstract: |
| An improved Particle Swarm Optimization (PSO) algorithm based on natural selection and nonlinear decreasing inertial weights is proposed to solve the problem of premature convergence and non convergence in PSO.In the process of algorithm iteration，the particle boundary velocity is limited by the nonlinear decreasing strategy of maximum velocity，and the nonlinear decreasing of inertia weight is used to balance the global search ability in the early stage and local optimization ability in the later stage.In order to keep the diversity of the population in the evolutionary process，the second order oscillation strategy is used in the standard particle swarm optimization to keep the diversity of the population in the evolutionary process.On this basis，the selection mechanism of genetic algorithm and particle swarm optimization are combined to improve the applicability of the algorithm.The proposed algorithm is verified by several benchmark functions and compared with other existing algorithms.Experimental results illustrated that this algorithm has more obvious advantages in search accuracy and optimization ability，especially in complex nonlinear optimization problems such as multi dimensional and multi peak optimization，the proposed algorithm has a strong competitiveness. |
| Key words: particle swarm optimization algorithm inertia weight natural selection maximum velocity nonlinear degression |
