Kalman Filter-based Fault Detection and Isolation of Direct Current Motor: Robustness and Applications

In recent years, as engineering systems have increased in complexity, aspects of reliability and safety are attracting more and more attention. Improved methods for detecting and isolating faults in engineering systems are of great practical importance. Traditional approaches to these methods have involved the limit checking of variables or applications of redundant sensors. More advanced methods have used residual analysis of signals by means of a comparing of actual plant behavior with the characteristics of a mathematical model. However, fault detection can be problematic because an "unknown uncertainty" is difficult to express in terms of a mathematical model. If these uncertainties can be considered, errors in fault detection and isolation in physical systems can be reduced. Therefore, this paper assesses a system's state after eliminating uncertainty with the kalman filter, and uses the experiment results to analyze and evaluate the performance of fault detection and isolation for a direct current motor.