Abstract: In order to solve the problems of the standard dung beetle optimization (DBO) algorithm, such as low convergence accuracy, lack of stability and easy to fall into local optimality, a multi-strategy improved dung beetle optimization (MSIDBO) algorithm was proposed. Firstly, the combination of Fun chaos and reverse learning strategy was used to initialize the dung beetle population, which increased the diversity and randomness of the population. Secondly, the global exploration strategy of the first stage of the osprey algorithm is introduced to replace the position update of the dung beetle rolling stage, make up for the fact that the dung beetle algorithm only relies on the worst value in the rolling stage, and accelerate the solving speed and accuracy of the algorithm. Thirdly, the integration of adaptive step size strategy and convex lens imaging strategy based on the updated foraging position of the small beetle has improved the algorithm's ability to develop globally and explore locally. Finally, adaptive t-distribution perturbation is applied to the foraging behavior of the thief beetle, enabling the algorithm to quickly jump out of local optima. The MSIDBO and other algorithms are tested on 14 functions. The results show that compared with other swarm intelligent optimization algorithms, MSIDBO's optimization ability and convergence ability are significantly higher than other algorithms. The improved algorithm is applied to the pressure spring design optimization problem, and it is further proved that the improved algorithm has better optimization performance.