An Algorithm for Parameter Identification of UAS from Flight Data

The aim of the present work is to realize an identification algorithm especially devoted to UAS (unmanned aerial systems). Because UAS employ low cost sensor, very high measurement noise has to be taken into account. Therefore, due to both modelling errors and atmospheric turbulence, noticeable system noise has also to be considered. To cope with both the measurement and system noise, the identification problem addressed in this work is solved by using the FEM (filter error method) approach. A nonlinear mathematical model of the subject aircraft longitudinal dynamics has been tuned up through semi-empirical methods, numerical simulations and ground tests. To take into account model nonlinearities, an EKF (extended Kalman filter) has been implemented to propagate the state. A procedure has been tuned up to determine either aircraft parameters or the process noise. It is noticeable that, because the system noise is treated as unknown parameter, it is possible to identify system affected by noticeable modelling errors. Therefore, the obtained values of process noise covariance matrix can be used to highlight system failure. The obtained results show that the algorithm requires a short computation time to determine aircraft parameter with noticeable precision by using low computation power. The present procedure could be employed to determine the system noise for various mechanical systems, since it is particularly devoted to systems which present dynamics that are difficult to model. Finally, the tuned up off-line EKF should be employed to on-line estimation of either state or unmeasurable inputs like atmospheric turbulence