The corrosion rate of grounding grid is an important aspect of corrosion state evaluation. The artificial intelligence algorithm model can predict the corrosion rate of the grounding grid well. In view of the problem that the selection of the characteristic input in the current prediction model is not comprehensive enough, based on the theoretical analysis of the grounding grid, the corrosion sampling point of the grounding grid is determined. The physical and chemical properties of the soil and the average growth rate of the grounding grid resistance are proposed as the characteristic input of the prediction model. The genetic algorithm (GA) is used to optimize the back propagation （BP） neural network, and the prediction model of the corrosion rate of the grounding grid is established. Compared with the unoptimized BP neural network model and the BP neural network model optimized by fruit fly optimization algorithm （FOA）, the prediction performance of the proposed model is better and has better applicability.