Predictive maintenance : utilizing AI-driven technologies to predict equipment failures on industrial facilities

Abstract

This study aims to provide a comprehensive evaluation of the AI-driven predictive maintenance approach, covering four key aspects of the topic through a systematic literature review: the motivators for adopting the approach, the key technologies that accompany AI-driven PdM, the models and methods with which AI-driven PdM predicts failures, and the operational and managerial impacts of its implementation in an industrial facility. From the research conducted, it was found that the adoption of PdM is mainly provoked not only by the pursuit of improvement, but also by the problems that the company or industrial facility either already has or will face in the future. There are various technologies that are implemented to enable and facilitate the PdM strategy, which are sensors, IoT solutions and digital twins. Machine learning algorithms are provided, which are usually used to classify data and its anomalies, if any, and these algorithms are used in modern AI PdM. Last but not least, there is an impact of the PdM implementation on the operational performance and management of a plant. All four research aspects contribute to a comprehensive understanding of the modern PdM approach, which may become even more widespread in the near future due to its long-term benefits.