In order to improve the search efficiency and convergence efficiency of the basic sand cat swarm algorithm, increase the diversity of the population, and enhance global search capabilities, an improved sand cat swarm optimization algorithm(IMSCSO) with multi-strategy mixing is proposed. Sine mapping is used for initialization to obtain a more evenly distributed population. In the attack behavior, the attack interval is divided according to the size of the individual adaptability to reduce the attack range and increase the search efficiency. The linear transformation of vectors is introduced into the search behavior, and the convergence efficiency is increased by the design of the coefficient matrix. The aggregation circle is used to increase the ability of the algorithm to jump out of the local optimum. Its ability of developing locally is enhanced with golden sinusoidal strategies which cite survival strategies. The improved algorithm is tested by 12 basic test functions, and the superiority of the improved strategy is verified by Wilcoxon rank sum detection, time complexity analysis, and Lyapunov stability analysis. Finally, the improved sand cat swarm algorithm is used to optimize the SVM parameters and applied to bearing fault detection, which proves the effectiveness of the algorithm in practical applications.