In order to improve the optimization efficiency of a competitive swarm optimization (CSO) algorithm, an improved triple competitive swarm optimization (ITCSO) algorithm is proposed for solving high-dimensional optimization problems. Firstly, a triple competition mechanism is used to improve the optimization efficiency of the algorithm. Simultaneously, the better convergence basis of the winners can improve the cognitive ability of the losers, and can guide the adaption direction of particles to improve the exploration ability. Secondly, the strategy that the losers with superiority fitness can learn from the winning subgroup is proposed, which can enhance the social cognition ability and reduce the number of evaluations, and can greatly improve the global search ability. Finally, the winning subgroup self-competition and the variation of losers with inferior fitness based on winners is proposed to enhance the local explore ability, which can avoid the algorithm falling into local optimum. In order to demonstrate the feasibility of the ITCSO algorithm, the stability and the convergence are proved by calculating the eigenvalues of the state transition matrix and using the limit analysis method. Several benchmark test functions are adopted to verify the performance of the proposed ITCSO. The experimental results show that, compared with other algorithms, the ITCSO not only has high optimization efficiency, but also has outstanding global search and local explore ability, which is more suitable for solving the high-dimensional problems. © 2024 Northeast University. All rights reserved.
