Comparative Analysis of Machine Learning Models for Predictive Analysis of Machine Failures

As we enter what most call the fourth industrial revolution, technological advancements have been made in terms of how equipment has been maintained. Predictive maintenance is a method that utilizes condition monitoring in order to prevent and predict machine and equipment failure. In recent years, machine learning has also been explored as a means to enhance predictive maintenance. In this study, the researchers created optimized models by cleaning and feature scaling the data then using logistic regression (LR), K-nearest neighbors (kNN), and artificial neural network (ANN) to create models that would predict and classify the failure of a machine given environmental and machine features as well as tool wear. Based on the confusion matrices and cross and hold-out validations conducted, ANN model proved to have the best results with it having the highest classification accuracy and its consistently accurate and balanced performance based on the different validation methods.