An Approach to Predicting the Oxidation of Coated Metals Through Machine Learning

The aim of the research is to predict weight gain of metal T11 during cyclic heating and cooling processes and by comparing their impact on different changes in additives. With the information obtained from which the base metal T11 was heated up at 800°C for an hour, and then left at room temperature for 20 minutes, with this process repeated for 50 cycles. Besides the base metal, three additive compositions are introduced in this work: NiCrAlY, NiCrAlY + Al₂O₃ and NiCrAlY + YSZ, with the same procedure applied for each composition. As such, it aims to build a machine learning model capable of predicting the weight change over future cycles from recorded data. Among the machine learning algorithms included, there are: Random Forest Regressor, Linear Regression, Gradient Boost, K-Nearest Neighbors (KNN) and Polynomial Regression. Metrics like Mean Squared Error (MSE) and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score show that the model with a Random Forest Regressor is the best predictor for all the metal compositions. This ability can then provide an interesting source to look at the applications at high temperatures through the long-term effects of thermal cycling on these materials.