This study explores the application of machine learning (ML) algorithms in predictive maintenance (PdM) within the manufacturing sector. The goal is to determine how different ML approaches optimally reduce equipment downtime by improving predictions on when failures are likely to happen in order to schedule maintenance. The study applies a design methodology comprising the conceptual design of a PdM system using sensors, and the performance evaluation of selected ML algorithms by Random Forests, Support Vector Machines, and Artificial Neural Networks was undertaken. Hypothetical results indicate that ML-based PdM is far superior to traditional maintenance approaches in terms of accuracy predicting equipment failures, greatly enhancing cost efficiency of maintenance services. This study brings forth the possibilities that machine learning brings to the maintenance of equipment in manufacturing providing higher efficacy and higher productivity.
Juma et al. (2024) studied this question.