Prediction of Raw Mill Fan Vibrations Based on Machine Learning Models

Abstract

Abstract In the cement manufacturing process, kiln process fans play a vital role. This article presents an extensive investigation into the prediction of Raw Mill Fan vibrations using machine learning models. The study evaluates multiple machine learning models, including k-nearest neighbors, linear regression, random forest, support vector machines (SVM), XGBoost, kernel ridge, support vector regression (SVR), recurrent neural networks (RNN), dense networks, and artificial neural networks (ANN). Performance evaluation metrics such as R2 and RMSE are employed to assess the accuracy of these models. The results demonstrate that both the Random Forest and XGBoost models exhibit remarkable predictive performance, with high R2 scores and low RMSE values. The conclusions drawn from these findings have important implications for the operation of the raw mill fan, as the adoption of predictive maintenance strategies can minimize downtime and enhance operational efficiency. Therefore, this study suggests that the incorporation of machine learning models into the monitoring and maintenance practices of kiln process fans has the potential to bring about a revolutionary improvement in the overall performance of the kiln process, yielding superior outcomes.