Predicting the electrical characteristics of oxyfuel cutting flame based on machine learning algorithms Available to Purchase

This research work discusses the development and application of predictive models based on machine learning (ML) algorithms to quantify the electrical characteristics of the oxyfuel preheat flame at different process states. By leveraging large datasets, ML can identify correlations between the measured electrical signals and the flame's thermal dynamics, which are difficult to model using traditional methods. Four ML techniques — Decision Trees (DT), random forest (RF), k-nearest neighbors (KNN), and Artificial Neural Network (ANN) — are utilized to predict the i-v characteristics of oxyfuel preheat flame exposed to an electric field, V ∈ [-10V, +10V]. The ML models were trained and tested using experimental data. Experimental measurements were performed using an Oxweld C-67 cutting torch fitted with a 12-inch two-piece fuel gas tip. The tip burns a premixed mixture of methane (CH₄) and oxygen (O₂) supplied through 12 preheat channels. The best fit ML model was identified using performance metrics such as R-squared (R₂), mean absolute error (MAE), mean square error (MSE), and root-mean-square error (RMSE). Among the models tested, the ANN model demonstrated the highest accuracy, providing predictions that closely aligned with the experimental results.