【摘要】 在扰动粒子群算法的基础上,提出了一种基于多样性策略的粒子群算法(ARPPSO)。该算法通过随机扰动全局极值对速度进行更新,并在速度更新中引入吸引和排斥机制控制种群多样性,同时在位置更新公式中,利用真实物理反弹理论将解空间外的粒子反弹回解空间内,有效地保持了种群的多样性。提出的算法经过基准函数的模拟实验验证,并与其他已有算法进行了比较。实验结果表明,除了保持粒子快速收敛性能外,很大程度上避免了早熟收敛,提高了收敛精度。特别是在复杂多峰函数优化上具有很强的竞争力。更多还原

【Abstract】 In this paper,based on perturbed particle swarm algorithm,a particle swarm optimization algorithm based on diversity strategy(ARPPSO) is developed.The new method introduces attraction and repulsion mechnisms into the velosity reform function,which is reformed by random perturbations global extremum;Considering that whichever particle moves out of the boundary in each dimension of solution space,the physical reflection theory is adopted to improve the efficient variety of the particle swarm.As a result,it effectively maintains the diversity of the population,through empirically testing and comparing with other published methods on benchmark functions.The experimental results illustrate that the proposed algorithm largely avoids premature convergence and improves convergence accuracy besides keeping good convergence performance.Especially it is very competitive for complex multimodal function optimization.更多还原