Non-linear aircraft flight path reconstruction review and new advances

Aircraft parameter identification techniques have become accepted as indispensable tools in the evaluation of prototype- and derivative aircraft in flight. Applications include estimation of stability- and control derivatives in the linearized equations of motion, synthesis of nonlinear aerodynamic and propulsion models in the context of performance certification and incorporation of information from dynamic flight test manoeuvres in a priori nonlinear flight simulation models. A variety of techniques in the time- as well as the frequency domain have been applied in the past. One of the successful techniques is the so-called two-step method (TSM) in which the original state-parameter estimation problem is decomposed into a nonlinear state/parameter estimation or reconstruction problem and a subsequent linear parameter identification problem. In the literature, the first step of the TSM is often referred to as `flight path reconstruction'. The present paper focuses on the first step of the TSM. After a derivation of the system models describing the flight path relative to a flat earth as well as a spherical and rotating earth, and observation models for air data and GPS, the flight path reconstruction problem is introduced. Requirements with respect to type and quality of flight test transducers are discussed. Next follows an overview of different approaches to the solution of the flight path reconstruction problem with emphasis on Kalman filter/smoother and Maximum Likelihood methods. A new adaptive algorithm is presented, the Modified Recursive Maximum Likelihood Adaptive Filter (MRML) which is shown to be significantly more robust with respect to initialisation errors than earlier methods. A reconstructibility analysis is presented for different transducer combinations. Numerical examples are presented based on simulated as well as actual flight test data. Flight results are given of the flight path reconstruction part of an on-line pseudo real-time application of the TSM. The paper ends with concluding remarks.