Pitting Corrosion Detection of Austenitic Stainless Steel EN 1.4404 in MgCl_2 solutions using a Machine Learning Approach

摘要

Different classification techniques such as Classification Tree (CT), Discriminant Analysis (DA), K-Nearest Neighbour (KNN) and Back-Propagation Neural Networks (BPNN) have been used to model pitting corrosion behaviour of austenitic stainless steel EN 1.4404. The main purpose is to predict the corrosion status of this material in different environmental conditions. Samples of this alloy have been subjected to polarization tests in order to determine pitting potentials values (Epit) with different aqueous conditions: chloride concentration (from MgCl_2 solutions), pH values and temperature. In this way, the classification methods employed try to simulate the relation between corrosion status and those various environmental parameters studied. These techniques have generally been regarded as successful, giving a good correlation between experimental and predicted data. High values for precision have been obtained for all the models making these techniques an useful tool to know the behaviour of austenitic stainless steel in different environmental conditions.