ABSTRACT
Machine learning is a method that allows interpretations of new data using an established database of older data by referring to already-known results (known as “labels”) and extrapolating between them to estimate the label that would be assigned to a different experiment. This can be a powerful tool for corrosion prediction, because it makes it possible to estimate a range of corrosion rates for a certain family of materials in a specific range of environments without actually performing experiments. In this paper, the machine learning concept was applied to the erosion-corrosion of steels in white liquor, a strongly alkaline industrial chemical used for pulping wood chips. Previously obtained corrosion data in white liquor, which included different steel compositions, particle concentrations and sizes, temperatures and fluid properties such as viscosity were compiled and assigned labels based on previous assessments in the industry as passive, acceptable, marginal or unsuitable according to observed corrosion rate. Models using thirty selected variables were built based on this data using diverse machine learning methods, including support vector machines (SVM), decision trees, k-nearest neighbor methods (KNN). discriminant analysis etc. Feature selection was attempted for each model. The best accuracies for each method were compiled and assessed regarding their promise for predictive purposes in erosion-corrosion
INTRODUCTION
Machine learning is a method that allows prediction of future data using an established database of older data by referring to classifications of already-known results, which form the so-called “training” dataset. The machine learning prediction is made by extrapolating between known results in order to estimate the label that would be assigned to a different experiment that is performed with largely but not completely similar parameters. This can be a powerful tool for corrosion prediction, because it makes it possible to estimate a range of corrosion rates for a certain family of materials in a specific range of environments without performing physical experiments. For the purposes of this paper, labels based on previous assessments of material performance in a pulp and paper environment were assigned to an original dataset relying on experimental data. A linear Support Vector Machine (SVM) was coded and applied in order to enable estimation and comparison of the corrosion performance of a variety of tested materials and environments that were not part of the original (training) dataset. Later, a publicly available MATLAB† application was utilized in order to test different methods of machine learning and their accuracies for prediction of labels.
