A Review of Phm System ’ S Architectural Frameworks

The need for a Prognostics and Health Management (PHM) system is propelled by increasing demand for condition based maintenance for systems to reduce cost of maintenance and mitigate risk. The use of multiple algorithms for PHM for a variety of systems presents challenges for PHM system developers in terms of integration and interfacing of various components, including hardware and software. A PHM system comprises of several elements including sensors, computing hardware and software algorithms for fault detection, diagnostics, prognostics and decision support. Thus there is a need for an architectural framework to help system developers and integrators for faster system development and deployment. An architectural framework provides a blueprint to enable the constituent subsystems to serve the overall purpose of the system. This also facilitates the interoperability of PHM systems with various applications. It provides a holistic view of the system for system developers and thereby forms the basis for building a system or serves as a guide for system modeling, integration and testing. This paper reviews the architectures proposed in literature for a PHM system. The architectural frameworks proposed for various PHM applications are also discussed and reviewed. The reviewed frameworks integrate various functionalities of the PHM system including data acquisition, signal processing, feature extraction, anomaly detection, diagnostics, prognostics and decision support. The architectures are categorized as either functional or physical architectures. The advantages and shortcomings of these proposed frameworks are also discussed.