Accelerometer Fault-Tolerant Model-Aided State Estimation for High-Altitude Long-Endurance UAV

This article proposes a novel fault-Tolerant dynamic model-Aided navigation filter to cope with accelerometer faults. An algorithm to estimate the three-Axis accelerations of a high-Altitude long-endurance (HALE) unmanned aerial vehicle (UAV) utilizing control input signals and aerodynamic coefficient parameters is newly proposed. To address the fault of the accelerometer, two model-Aided navigation filters that utilize the measured acceleration, denoted as Acc-measure algorithm, and estimated acceleration, denoted as Acc-free algorithm, respectively, are effectively combined under the interacting multiple model (IMM) framework to integrate the optimality of Acc-measure algorithm and robustness of Acc-free algorithm. Flight test results demonstrated that the proposed algorithm yields robust attitude and wind estimation results in the presence of different types of accelerometer faults compared with Acc-measure and Acc-free algorithms while accurately detecting the fault of the accelerometer.