Sliding mode output estimator based fault detection, isolation and estimation for systems with unmatched unknown inputs

This paper considers fault detection, isolation and estimation problems for a class of systems with unknown inputs which may not satisfy certain matching conditions. Such conditions are often required for existence of sliding mode observers (SMO). In cases when the unmatched unknown inputs are present, no SMO could be designed such that the state estimation error is invariant to all the unknown inputs, and therefore, existing SMO based fault diagnosis schemes can not be employed. In this article we propose a novel approach to design of output estimators using sliding mode technique. The estimators are then used for fault detection, isolation, and estimation. First, a canonical representation of the system which decouples the matched and unmatched unknown inputs is derived. Second, based on this canonical form, output estimators using sliding mode technique are proposed, and their properties are investigated. Third, a fault diagnosis scheme is developed to carry out the fault detection, isolation, and estimation tasks. Finally, an example is given to show the effectiveness of the output estimator based fault diagnosis scheme in terms of fault detection, isolation and estimation.