An Unknown Input Observer-EFIR Combined Estimator for Electrohydraulic Actuator in Sensor Fault-Tolerant Control Application

This article presents a novel unknown input observer (UIO)-integrated extended finite impulse response (EFIR) estimator and its application for an effective sensor fault-tolerant control (FTC) of an electrohydraulic actuator (EHA). The proposed estimator exploits the UIO structure in the EFIR filter. Thus, it requires only a small amount of historical data (N) while ensuring the following: 1) sensor fault and system state estimation accuracy under time-correlated noise; 2) the number of estimator design parameters is significantly minimized; and 3) robust residual generation. A Lyapunov-stability-based theory is carried out to study its convergence condition. Next, an EHA-based test rig has been set up, and sensor FTC is performed by carrying this estimator as part of a fault diagnosis algorithm to evaluate its performance by both simulation and real-time experiments. Results highlight that under optimal setting (N = N-opt), the estimator performance is near accurate to the very well developed extended-Kalman-filter-based UIO in undisturbed conditions but significantly outperforms when dealing with time-correlated noise under the same control environment. The estimator also shows its robustness under below-optimal setting (downgrading N-opt by 50%) while performing sensor FTC in real time.