An Improved Sparrow Approach

Abstract

The basic sparrow search algorithm reduces the population diversity at the later stage of the iteration and is easy to fall into local extremes. An improved sparrow algorithm (ISSA) is proposed basen on fusion of Cauchy mutation and reverse learning. Firstly, a Sin chaotic initialization population with unlimited number of mapping folds is used to lay the foundation for global optimization. Secondly, the previous generation global optimal solution in the discoverer position update method is introduced to improve the sufficiency of the global search, and at the same time, adaptive weights are added, local mining and global exploration capabilities are coordinated to speed up the convergence speed; then, the Cauchy mutation operator and the reverse learning strategy are combined to perform perturbation mutation at the optimal solution position, new solutions are generated, the algorithm's ability is enhanced to jump out of local space. Finally, the paper compared with three basic algorithms and two improved sparrow algorithms.