Using neural network techniques to predict crack growth rates of stress corrosion

An artificial neural network (ANN) with the learning rule of a stochastic process has been developed to describe intergranular stress corrosion cracking (IGSCC) in sensitized Type 304 stainless steel in high temperature aqueous solutions. The ANN predictions of crack growth rate (CGR) versus oxygen concentration, flow velocity, stress intensity, hydrogen concentration and ECP have been successfully presented. Finally, the steady state crack growth rate predicted by the ANN at low ECP values has been verified by Wilkinson's theory.