Abstract: To remedy the defects of particle swarm algorithms, including the local optimum, low convergence accuracy, and slow convergence speed, this paper proposes an improved particle swarm algorithm based on hybrid strategies. First, the population is initialized by the fusion Circle mapping and the elite reverse learning to improve its quality and accelerate the convergence. Second, the spider mobile strategy is introduced in the particle speed update to balance the local and global search of the algorithm; then self-adaptive T distribution is proposed to enhance the algorithm’s global search and its ability to jump out of local optimum. Finally, the 15 single-peak and multi-peak functions are simulated and analyzed with the other three algorithms. Our results show the improved algorithm possesses strong optimizing capacity and stability.