An actuator fault isolation strategy for linear and nonlinear systems

This paper investigates actuator fault isolation problem for linear systems. Then, as an extension, actuator fault isolation problem for nonlinear systems is considered next For both linear and nonlinear systems, two cases are studied. In the first case, we assume that all states are available. The second case assumes that only outputs are available. To accomplish fault isolation, we use a bank of observers for all possible faulty models. The paper considers constant actuator faults, that is, the outputs of some actuators are stuck at fixed undesirable constant values. The fault isolation strategy is based on combining the conventional observer design techniques with adaptation techniques. Based on the designed observers, a bank of residuals are defined correspondingly. The actuator faults can be isolated if only one residual goes to zero while the others do not. The faulty model with residual approaching zero identifies the faulty actuators. For linear and nonlinear systems with all states available, new sufficient conditions for fault isolation are derived, which require only that the distribution matrices of the actuators are of full column rank. For linear and nonlinear systems with only outputs available, new sufficient conditions for fault isolation are also derived, which require additional conditions besides the full column rank condition of the distribution matrices of the actuators. Some simulation results are given to show the effectiveness of the proposed fault isolation methods.