Aiming at the problem that the reliability and combat effectiveness of equipment are reduced due to corrosion, a corrosion rate prediction model for equipment components based on kernel extreme learning machine (KELM) is proposed. The kernel principal component analysis (KPCA) was used to preprocess the original data, and then the improved hybrid dung beetle optimization (HDBO) algorithm was used to optimize the kernel parameters and regularization coefficients in KELM, and the Singer chaotic map was introduced to improve the initial distribution position of dung beetle population; A variable spiral search strategy was used to improve the foraging position update process of the spawning dung beetle and foraging dung beetle, and to expand the ability of the beetle to explore the unknown area.Levy flight strategy and adaptive weight were introduced to enhance the global search and local optimization ability of the algorithm. The corrosion rate prediction model of KPCA-HDBO-KELM equipment was established. Sixty groups of experimental data of corrosion rate of equipment metal structures were used to verify and compare with the other three models. The results show that the KPCA-HDBO-KELM model is more accurate and stable than other models, and the MSE and MAE indicators are better than comparison model, which has high stability and accuracy in predicting the corrosion rate of equipment components.