The grounding grid is an important component of the substation, and its corrosion will directly influence the safe operation of the power system. In order to overcome the limitations of traditional corrosion diagnosis methods, the corresponding corrosion diagnosis equations model was constructed by using electric network theory. Considering the limited number of measurable nodes in the actual substation, which leads to the characteristics of high-dimensional underdetermination of the equations, an improved RMSProp-Lookahead hybrid optimization algorithm was proposed to solve the problem. The method combined the Gradient normalization mechanism and the Lookahead dual update strategy, optimized the hyperparameter configuration through grid search, and improved the convergence speed and global search capability of the optimization process. The simulation and experimental results showed that the method can effectively locate the corrosion branch, accurately judge the corrosion degree, and significantly improved the diagnostic efficiency and accuracy, which has important practical significance for the safety maintenance of the grounding network.