Robust fault-tolerant control using an accurate emulator-based identification technique

A robust fault-tolerant control scheme using an accurate and robustly identified model of a system operating in a closed loop is proposed. A robust identification of the system and the associated Kalman filter is proposed by the novel use of an emulator, which is a transfer function connected to the input, output or both, to mimic the likely operating scenarios, both normal and abnormal. From the emulator-generated data, a set of perturbed models are generated, which plays a crucial role in ensuring robustness of the identified model and subsequently that of the controller. The prediction error method is employed to identify the perturbed models and a robust optimal model is obtained as a best fit to the perturbed models. A robust Kalman filter-based state feedback controller is obtained from the parameterisation of all stabilising controllers generated using the emulator-perturbed models. It is shown theoretically that the proposed robust controller designed using the emulator-perturbed robustly identified model is superior to the conventional robust controller designed from the identified nominal model, and a significant improvement in robustness is confirmed using illustrative examples. A fault-tolerant control system is developed by exploiting the key properties of the freely available Kalman filter.