Predictive Maintenance in Manufacturing via Machine Learning Algorithms

Abstract

Purpose:

The goal of this project was to investigate machine learning methods for industrial predictive maintenance. approach: A desktop research approach was used in this study. Secondary data, or data that may be gathered without fieldwork, is referred to as desk research. Since desk research mostly entails gathering data from already-existing resources—executives' time, phone bills, and directories—it is frequently seen as a less expensive method than field research. As a result, the study used data, reports, and studies that have already been published. It was simple to obtain this secondary data by using the library and internet journals.

Findings:

The results indicate that there is a methodological and contextual gap concerning machine learning algorithms for predictive maintenance in manufacturing. A preliminary empirical evaluation found that using cutting-edge machine learning methods significantly increased the efficacy of predictive maintenance plans. The study showed that complex models with higher accuracy in equipment failure prediction and maintenance schedule optimisation were deep learning and ensemble approaches. The significance of real-time monitoring and high-quality data for improving predictive skills was also emphasised. The study found that, despite these developments, there are still issues with computing capacity and implementation complexity. These issues need to be resolved in order to fully realise the advantages of machine learning technologies in manufacturing.

Unique Contribution to Theory, Practice and Policy:

Machine learning algorithms for predictive maintenance in manufacturing may be investigated using the theories of predictive analytics, machine learning classification, and anomaly detection. The research made a number of recommendations for improving the predictive maintenance use of machine learning techniques. It suggested that practitioners tackle issues like computing needs and model complexity, embrace sophisticated algorithms like Neural Networks and ensemble approaches, and make investments in high-quality data collecting. It recommended creating frameworks to assist in the efficient application of these technologies while attending to cybersecurity and data privacy issues for regulators. The study also made clear how important it is to carry out further research on organisational issues, technological integration, and hybrid techniques in order to enhance the results of predictive maintenance and propel the area forward.