PREDICTIVE MAINTENANCE TO REDUCE MACHINE DOWNTIME IN FACTORIES USING MACHINE LEARNING ALGORITHMS

Abstract

Accurate machine failure detection allows manufacturers to estimate potential machine deterioration and avoid machine downtime caused by unexpected performance issues. Predictive maintenance with the use of machine learning algorithms may anticipate machine faults and maximize maintenance efforts to solve machine downtime problems. To anticipate machine breakdowns and minimize downtime, this work applies a variety of machine learning methods, such as Random Forest, Decision Tree, Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Gradient Boosting, and Logistic Regression. Based on the performance measurement values, Random Forest model has shown high levels of accuracy, precision, recall, and F-score. The sequence of order for accuracy of machine learning models follows as: Random Forest > Decision Tree> Gradient Booster Classifier and SVM > Logistic Regression and KVM. This work emphasizes that, through various machine learning models, machine manufacturers could optimize the machine maintenance and prolong the life of machines.