Integrated Modeling for Remaining Useful Life Prediction and System Health Management

Description

Reliability assessment plays an important role in engineering design and maintenance management. Current methods for reliability assessment of complex systems have fundamental flaws, due to their inability to keep pace with new technologies, and to account for complex usage profiles. Failures of systems are common due to their complexity (e.g., airplane crashes caused by failed electronics, power grid shutdowns caused by failed sensors, and financial shut-downs due to failed servers). The impact of these failures to the society on safety, availability, and cost is staggering. Newly improved modeling techniques are needed for reliability and degradation assessment, fault diagnostics, and prognostics (the real-time prediction of reliability and the remaining useful life) of complex systems. The proposed research approach is a radically new approach, which focuses on effective and efficient reliability prognostics and system health management (PHM) for complex systems based on integration of failure time and degradation data, physics of failure knowledge, and the information on the actual field operational conditions of the systems. The proposed research methods can be applied to a wide range of complex systems, including electronic-rich systems, critical automotive components, and power systems. In particular, we will investigate its application to the rechargeable batteries, a widely used power sources in many electrical and electronic systems today. There is a strong industry demand both in Hong Kong and China to raise quality and reliability to a new international level. The proposed PHM technologies and remaining useful life (RUL) estimation methods will meet this challenge. The main objective of this research is to develop an integrated modeling approach for accurately predicting system health that helps engineers and scientists understand and quantify the impact of risk and uncertainty in making reliability and maintenance decisions. In particular, we will (i) develop new classes of integrated models that combine failure time and degradation data together with dynamic usage information to predict malfunctions and remaining useful life for complex systems, with a particular emphasis on battery systems; (iii) develop model validation metrics and test cases for validating and assessing the proposed methods.