NEURAL NETWORK APPROACH IN RESEARCH OF METAL CORROSION IN SOIL
神经网络在金属土壤腐蚀研究中的应用

This paper introduce neural network approach to study the non-linear reationshipbetween the physical and chemical properties and corrosion rate of carbon steel in soil from the collected corrosion data. The corrosion rate of carbon steel in soil can be predicted by trained neural nerwork. In the study, a neural network with 5-8-1 structure was used. The learning algorithm is BP (back-propagation) algorithm. As a result, H2O and Cl-ion in soil are dominant factors influencing corrosion of carbon steel.