Airflow Angle Estimation for Large Civil Aircraft under Non-Uniform Observability Condition

This paper investigates the estimation of angle of attack (AOA) and angle of sideslip (AOS) using aircraft kinematics. The proposed estimator relies on an approximate linear time-varying kinematic model, and utilizes measurements from the GPS, inertial measurement unit, and pitot tube. Although the entire state vector is shown to be unobservable in the presence of a constant flight attitude, we perform a novel analysis on observability degree by evaluating the angle between the direction vector of each state variable and the observable subspace, and reveal that both AOA and AOS have a high observability degree as long as the airspeed is sufficiently large. To address the ill-conditioned problem associated with the above non-uniform observability condition, a reduced-order Schmidt-Kalman filter (SKF) based method is proposed to estimate only the strongly observable AOA and AOS in the state vector. The unobservable wind components are not explicitly estimated, but their influences are still considered for updating the covariance matrices and the filter gain. Using real cruise flight data of a large civil aircraft, the proposed SKF based method is demonstrated to have about 40% reduction in root mean square error in comparison with using the standard Kalman filter under the non-uniform observability condition.