Adaptive Sensor Fault Detection and Isolation in Uncertain Systems

Abstract

An adaptive sensor fault detection and isolation approach in linear multi-input multi-output (MIMO) systems with unknown system parameters is presented. The proposed diagnostic approach abandons the idea of designing adaptive observers to estimate the system's state, and rather employs the design of adaptive output estimators for estimating only the outputs. First, a MIMO system is decomposed into a group of MISO systems and a transfer function description for each MISO system is presented. Second, based on each transfer function and for each output, an output equation, which is suitable for output estimator design, is obtained by filtering the corresponding output and all the inputs properly. Third, using the derived output equations, adaptive output estimators are designed for all outputs. Finally, based on the designed output estimators, the adaptive sensor fault diagnosis problems are solved. The proposed fault diagnosis scheme enables us to treat each output separately, and this turns the difficult sensor fault isolation problem into a much simpler task. Another advantage offered by the proposed approach is that it does not require the original systems to be detectable. The results presented in this respect are even new for known linear MIMO systems because no such scheme has been proposed in the literature in the past. A linearized aircraft model is used as an example to show the effectiveness of the output estimator based fault diagnosis scheme in terms of sensor fault detection and isolation.