Abstract: Aiming at the problems of slow convergence speed, poor optimization accuracy and easy to fall into local optimal solution of the coati optimization algorithm, a multi-strategy improved coati optimization algorithm is proposed. Refraction reverse learning is introduced to initialize the coati population, which ensures the diversity and uniform ergodicity of the population distribution, and improves the convergence speed and optimization accuracy of the algorithm. The Levy flight strategy is introduced to improve the exploration phase of the coati optimization algorithm, which enhances the ability of the algorithm to jump out of local optimal solution. Inspired by the whale optimization algorithm, the spiral search mechanism is introduced into the development stage of the optimization algorithm, which enhances the global search capability and local exploration capability of the algorithm. The adaptive t-distribution mutation iteration method is introduced, which balances the global search capability and local development capability of the algorithm. Comparing with various intelligent algorithms, the good performance of the improved algorithm is verified through 12 multi-type test functions. In order to further evaluate the effectiveness of the improved algorithm, it is used for the optimization of extreme gradient boosting model parameters. The experimental results show that compared with the other five algorithms, the improved algorithm has higher classification accuracy and convergence speed.

Key words: coati optimization algorithm, Levy flight strategy, spiral search mechanism, refraction reverse learning, adaptive t distribution mutation