Enhancing sand cat swarm optimization based on multi-strategy mixing for solving engineering optimization problems

The Sand Cat Swarm Optimization (SCSO) algorithm is characterized by slow convergence speed and a tendency to become trapped in local optima. To address these issues, a hybrid multi-strategy variant known as HSCSO is proposed. The algorithm first employs an Elite Opposition-Based Learning strategy, which generates complementary candidate solutions, providing a more diverse set of initial solutions and improving the distribution of individual positions. Subsequently, a greedy strategy is utilized to combine the position update strategy of the golden sine algorithm, which has a fixed search pattern, with the position update strategy of the SCSO, which features random search characteristics. This combination retains the advantage of fast convergence while also providing a diverse set of random candidate solutions. Finally, a Random Inertia Weight is introduced to assist the SCSO algorithm in adaptively adjusting the search step size and optimization direction based on the different phases of the search process, thereby helping the algorithm effectively avoid local optima. The improved algorithm was compared with two classic original algorithms and six different algorithm variants on 25 benchmark functions. The experimental results indicate that HSCSO exhibits faster convergence and higher solution accuracy on most benchmark functions. Furthermore, the performance of HSCSO was evaluated on the CEC2019 and CEC2021 test suites, as well as several engineering optimization problems. The experimental results validate the competitiveness and superiority of the proposed HSCSO algorithm, demonstrating its potential for application in the field of global optimization.