Prediction of the remaining useful life of a milling machine using machine learning

The cutting tool is a key component of the milling machine that decides productivity. Hence, an adequate predictive maintenance (PdM) strategy for the cutting tools becomes necessary. This research seeks to develop a smart maintenance web application that utilizes Machine Learning (ML) supervised models to predict the Remaining Useful Life (RUL) for milling operations. The ML models were developed using a four-stage process including data pre-processing, training, evaluation, and deployment. Several ML algorithms were applied and the results were evaluated using five measures involving Accuracy, Mean Absolute Error (MAE), Mean Squared Error (MSE), R-squared, and R-squared adjusted. It was found that the Multi-Layer Perceptron Regressor provided the largest accuracies, adjusted R-squared, MAE, and MSE of 99 %, 0.99, 3.7, and 23.13, respectively. A web application for maintenance was finally developed with several ML algorithms at the evaluation stage. Maintenance engineers can utilize the developed smart web application to monitor the machine's health state and predict failure occurrence. In conclusion, the developed web application assists engineers in developing reliable predictions of maintenance activities, which may save costly production and maintenance losses.

• •
  A Web application based on machine learning techniques was developed for RUL predictions for the milling cutting tool.
• •
  A comparison between the prediction results from various machine learning techniques was conducted.
• •
  The web application is found to be valuable for maintenance prediction and planning.