Fault-Tolerant Active Vibration Control

Abstract

The objective of this work is to detect failures in a vibration control system and adapt the control system in order to maintain optimal performance. Fault detection and isolation (FDI) filters, which are a subset of state observers specifically designed to detect and identify known types of system failures, are used to detect system failures. The various types of system failures considered in this work are sensor and actuator malfunctions (additive failures), and so called parametric failures(multiplicative failures). Two types of FDI filters are studied in the current research work: parity relations and Beard-Jones (BJ) filters for additive failures. Similarly, parameter estimation methods and BJ filters are studied for multiplicative failures. The output of such filters is used, along with hybrid automata, to reconfigure feedback compensators in order to maintain closed loop objectives. Such reconfiguration allows the system to continue operating optimally under certain, pre-defined system failures. The two types of FDI filters are compared in order to establish the appropriate filter design for experimental work. Therefore, a new methodology is developed for BJ filter design for higher order systems and, also certain classes of systems that resemble the anticipated real-time systems. The primary contribution of this work is the integration of active control with fault detection and hybrid system management to achieve fault-tolerant control. A second contribution is new BJ design technique for real-time use. The third contribution is experimental demonstration of fault-tolerant active vibration control. There are several challenges inherent in this effort but the most important is the management of compensator switching. Since switching involves system discontinuity, stability of the system is very difficult to guarantee. The various results demonstrate some of these challenges and the effectiveness of BJ FDI filters in maintaining suitable closed loop performance in the presence of system malfunctions.