基于动态预警与选择变异的改进麻雀分类算法

Improved sparrow classification algorithm based on dynamic early warning and selected variation

针对麻雀搜索算法寻优性能低、多样性不足问题,为提高算法的全局搜索能力和局部开发能力,提出一种基于动态预警与选择变异的麻雀搜索算法(CGSSA).采用线性微分递减的方式动态设定优化过程中的预警者数量,在更新过程中设定选择因子,按照震荡衰减的规律对算法中一定数量的较差个体进行变异,并在变异结束后对更新前后的个体择优选取.实验结果表明:CGSSA算法不仅在收敛速度、搜索精度以及稳定性上具有更好的优化性能,与BP神经网络相结合应用于机器学习中的分类任务时,具有更好的分类性和鲁棒性.

In order to improve the global search ability and local development ability of the Sparrow Search Algorithm,Coyote Genetic Sparrow Search Algorithm(CGSSA)is proposed for the shortcomings of low optimization performance and insufficient diversity.The number of predictors in the optimization process is dynamically set by linear differential decline method.Then the selection factor is set after the algorithm update process,and a certain number of poor individuals in the algorithm are mutated according to the law of shock attenuation,and the individuals before and after the mutation are selected.The experimental results show that the CGSSA algorithm not only has better optimization performance in terms of convergence speed,search accuracy and stability,but also has better classification and robustness when it is combined with the BP neural network and applied to the classification task in machine learning.