Digital Twins in Prognostics and Health Management: A Review of Diagnostic, Remaining Useful Life, and Predictive Maintenance Applications

Prognostics and Health Management (PHM) systems continuously monitor industrial operations and component health to detect unusual patterns and reduce downtime failure rates. Several PHM reviews have been published in recent years, offering comprehensive overviews of the latest approaches and methods. Nowadays, the digital representation of physical machines and their operations has become a popular practice for predictive maintenance. Thus, Digital Twins can be used to simulate real-world systems and outcomes in nearly real-time, allowing us to make better decisions. However, there has been insufficient focus on diagnosis and Remaining Useful Life (RUL) management aspects using Digital Twins in terms of predictive maintenance. In addition, the lack of information on the management of operating conditions is also a prominent issue. How can Digital Twins be used in creating a virtual environment for operations and being aware of upcoming conditions before they happen? To address these gaps, with this article, we intend to present an overview from three different points of view: diagnosis, RUL, and operating conditions. We exhaustively review relevant and impactful publications, where these three viewpoints relate to predictive maintenance. This review aims to provide an overview and identify gaps in the use of digital twins in PHM, extracting relevant guidelines for future developments.