Abstract / Description

This paper describes a frequency-domain greybox system identification procedure for scale model helicopters. It presents a unique way to obtain the initial constraints of the unknown parameters and with this additional information, the identification becomes more efficient. There is no need anymore to start the estimation from blind initial guess or get good starting values through tedious experiments. A well-known 6- DoF LTI MIMO state-space model suitable for hover and near hover condition is used in this paper. In order to identify all the unknown parameters, the model is partitioned into several subsystems in a systematic way. The performance of the proposed procedure is tested in simulations using corrupted flight data by comparing the identified parameters, the eigenvalues and the predicted outputs to their `true` values, respectively.