Cong Wu, Department of Applied Mathematics, University of Waterloo
Model-based Fault Estimation
Many control systems in industry are required to be safe, effective and reliable. Faults in such systems can cause malfunction, failure and even disaster. A traditional approach to detect fault and appraise effects of fault in a system is based on direct measurement of fault. Since some faults and variables cannot be measured directly, it is necessary to develop dynamical models using other measurable variables to estimate them. In this talk, we introduce model-based fault estimation for control systems, where observers of fault and feature variables are designed to monitor the target system when it is fault-free and to report an estimate with a bounded error when it is faulty. We will discuss some fault models, give examples to illustrate the motivation of fault estimation, introduce the process of fault estimation and propose problems for future work.