Projection-Based Online Parameter Estimation of a Tilt-Rotor VTOL Aircraft and Experimental Validation

Abstract

Abstract. The vertical take-off and landing (VTOL) aircraft exhibit complex and rapidly varying dynamics during the transition flight phase, which poses significant challenges for accurate modeling and aerodynamic parameter estimation. This article presents a parameter estimation framework to identify unknown aerodynamic coefficients governing the transition flight, leveraging experimental flight data of a tilt-rotor VTOL unmanned aerial vehicle (UAV) in outdoor experiments. The VTOL UAV has a hybrid configuration with two tilting rotors, two static rotors, and fixed wings. A nonlinear dynamic model describing the longitudinal motion is developed and reformulated into a regression structure suitable for parameter estimation. A projection-based constrained recursive least-squares algorithm is then applied to estimate critical aerodynamic parameters, including lift, drag, and thrust coefficients, under physical constraints. The convergence and accuracy of estimation algorithm for time-varying coefficients are first verified by simulation. The parameter estimation is further validated by six experimental flight tests with different tilting rates of 12 deg/s and 14 deg/s. Experimental results demonstrate the accurate estimation by accurate predictions of the aircraft states vx, vz, ωy with root mean square error of 0.7230 m/s, 0.0990 m/s, and 0.0508 rad/s, respectively.