Application of Neural Network in Corrosion Property of High Speed Steel

The article is dedicated to the application of neural network in corrosion property of high Speed Steel. The corrosion properties of high speed steel with about 9wt% vanadium content and different carbon content were tested under different H3PO4 medium concentration conditions. Using back-propagation (BP) neural network, the non-linear relationship model among the corrosion weight losses (W), corrosion parameters (corrosion time, H3PO4 concentration) and alloy composition (carbon content) is established according to the tested experimental data. The results show that the neural network model can predict the corrosion weight loss precisely according to corrosion conditions and alloy composition. The prediction results reveal that the corrosion resistance of high speed steel decrease with the increase of H3PO4 concentration or carbon content in high speed steel. It is suggested that the corrosion condition and alloy composition should be considered synthetically to estimate the corrosion property of high speed steel.