基于动态多子族群自适应群居蜘蛛优化算法

An Adaptation Social Spider Optimization Algorithm Based on Dynamic Multi-swarm Strategy

摘要:

为了提高群居蜘蛛优化算法（SSO）样本多样性和算法收敛性能，提出了一种基于动态多子族群自适应群居蜘蛛优化算法（DMASSO）。根据算法样本多样性和算法进化程度，动态的将蜘蛛种群分成若干个主导子族群和辅助子族群，在不同子族群中分别引入自适应学习因子和高斯扰动因子改进算法个体更新方式，实现提高算法全局寻优能力和保持群体样本多样性。针对具有典型特点的测试函数仿真结果表明，较SSO算法、MSFLA算法等优化算法相比，新算法在收敛速度和收敛精度上均有明显改善。

Abstract:

In order to improve the samples diversity and convergence properties of social spiders optimization algorithm (SSO), an adaptation social spider optimization algorithm based on dynamic multi-swarm strategy (DMASSO) is proposed. According to the algorithm samples diversity and evolutionary level, the spider population is dynamically divided into different sizes leading groups and supporting groups, and the adaptive learning factor and Gaussian disturbance factor are introduced to improve the algorithm update ways, which helps to improve the algorithm global optimization ability and maintain the diversity of the sample population. For the test results of typical characteristics functions show that compared to SSO algorithm, SFLA algorithm and other optimization algorithms, the new algorithm has better convergence speed and convergence accuracy.