Study on prediction of soil corrosion in power grounding based on artificial neural network

The soil corrosion prediction of grounding material provides an important basis for power design and management. At present, grounding material corrosion is mostly assessed by single factor or multi-factor comprehensive evaluation without giving accurate quantitative data. In this paper, the key factors of soil corrosion were studied and analyzed, and the BP neural network method was used to train the key parameters of the transmission tower grounding material and the soil electrochemical corrosion rate data. An artificial neural network was established by using 8 characteristic parameters and soil corrosion rate data measured in Guizhou Province, and a method for accurately predicting soil corrosion rate of grounding materials was found through training. The study found that parameters such as soil resistivity, water content, salinity, pH value, self-corrosion potential, redox potential, chloride ion, sulfate ion had a good correlation with soil corrosion rate. The research results showed that the method could accurately calculate the soil corrosion rate of the grounding material and provide accurate data support for the management of grounding material.