Application of Second Order Sliding Mode Observers for Fault Reconstruction on the ADDSAFE Benchmark

Abstract

This paper presents applications of second order sliding mode observer schemes to two different ADDSAFE aircraft fault detection benchmark problems. Firstly, the detection and isolation problem associated with a faulty yaw rate sensor is considered and secondly an oscillatory failure case is tackled. For the yaw rate sensor fault problem, a robust sliding mode reconstruction scheme for the aircraft lateral axis based on an LPV system is presented. The scheme involves transforming the sensor fault reconstruction problem into an actuator reconstruction problem. The fault reconstruction is obtained from the equivalent output error injection signals associated with a second order sliding mode structure which is less susceptible to noise due to its natural filtering properties compared to the typical signum function associated with first order sliding. For the OFC problem, this paper proposes a nonlinear scheme for reconstructing the OFC using a nonlinear model of the hydraulic actuator. The reconstruction scheme requires an estimate of rod speed provided by a second order sliding mode observer. Ideally low gains in the observer are required because of the noisy environment associated with the physical system. However during high frequency and high amplitude OFCs, the observer gains are allowed to adapt to maintain sliding. Simulation results from the high fidelity nonlinear aircraft model (which includes sensor and process noise) are presented for both the sensor fault and the OFC. The results show good fault reconstruction for a wide range of flight conditions for yaw rate sensor faults and different amplitude and frequency OFCs (both liquid and solid).