Parameter identification

Page 1

FLIGHT DYNAMICS, TOPIC 5 PARAMETER IDENTIFICATION 5.1 Methods for system identification A problem of model identification consists in the ability to extrapolate a mathematical model from experimental data referred to a dynamic system. If the structure of the model is known, the problem is termed as parameter identification, and only the coefficients of the state matrices might be derived (such as aerodynamic and control derivatives for the flight dynamics case). As far as the flight dynamics concerns, the structure of the model has already been given, hence some parameter identification techniques should be implemented in order to fill the state matrices. As a very first approach, the recurring to well-validated historical data referred to aircrafts of similar configuration and mission. This is a fist attempt of parameter identification that gives at least significant results in the design phase. Of course, if a very innovative architecture is designed, the historical data will not provide good parameters to run the simulation model. The further phase is the wind tunnel testing, which provides aerodynamic data that approach the reality, from which the parameters can be deduced. However, performing scaled experiments comports an almost not negligible distance between the wind tunnel data and the full scale model data, since Mach, Froude and Reynolds numbers can not be reproduced correctly. Of course the wind tunnel allows to run the simulation model and, as a consequence, design corrections are possible. Anyway, if the purpose is to implement a training simulator or to design a control system, the simulation model must completely match the real dynamics. Thus, the parameter identification is provided by using flight test as inputs data. If the simulation model is forced…