Retrospective Cost Model Refinement for On-Line Estimation of Constant and Time-Varying Flight Parameters

In this paper we apply retrospective cost model refinement (RCMR) to parameter estimation. To gain insight into the accuracy and speed of convergence of the RCMR estimates, we consider aircraft dynamics with uncertain entries in the dynamics and input matrices. We consider scenarios that include multiple uncertain parameters, alternative measurements, and sensor noise. All of the examples are discrete time, as required by RCMR. To account for the structure of the uncertain parameters, we develop a Jacobianbased technique for constructing the appropriate feedback structure. We use this approach to account for the appearance of stability derivatives in the dynamics matrix, and we use this technique to estimate airspeed variations in the vicinity of trimmed flight.