A multi strategy improved sand cat swarm optimization algorithm

https://elibrary.ru/SEDEZT

Abstract

In order to overcome the shortcomings of the development capability of the standard sand cat swarm optimization algorithm and avoid its tendency to fall into local optima, this paper proposes an improved sand cat swarm optimization algorithm (MSISCSO) that integrates multiple strategies. Firstly, use Sin Tent Cos chaotic mapping to initialize the initial individuals of the sand cat population and increase the diversity of the initial population; Secondly, the first stage global development strategy of the Raccoon Optimization Algorithm is introduced to update the position of the sand cat, enhancing its convergence speed during the development stage; Furthermore, introducing a crossover strategy instead of the position update formula during the exploration phase of the sand cat group enhances its ability to escape from local optima; Finally, the sand cat group is subjected to adaptive t-distribution perturbation to further enhance its global exploration ability. Comparative tests were conducted on six test functions using MSISCSO and five other algorithms, and the results showed that the MSISCSO algorithm had faster convergence speed and accuracy on all six test functions. Finally, the MSISCSO algorithm was applied to solve spring design problems, demonstrating that the improved algorithm can solve complex engineering optimization problems.