عنوان مقاله [English]
Most of the Unmanned Aerial Vehicles (UAVs) fly at low speed and low Reynolds number regimes and may have complex nonlinear dynamics. Hence, using available aerodynamic and propulsion software is not always reliable. On the other hand, dynamic modeling of UAVs by experimental test data is expensive and time consuming. For this reason, dynamics identification is a simple and useful solution. In the current study, the dynamic model of a fixed-wing UAV has been identified using flight test data. The UAV has a canard configuration and is powered by an internal combustion engine with a fixed pitch propeller. Because of longitudinal and lateral dynamic coupling, due to rotary propeller and polyhedral wing, a multi-input multi-output (MIMO) dynamic model in the state space form has been identified. The identification process has been performed by two scenarios. At first, dynamics is identified in a way that it can be used for investigation of stability as well as designing MIMO controllers. This model can also be used in online investigation of instrument degradation and faults. The aim of the next scenario is to identify the dynamic model which can be used in software simulation. The identified dynamic models are also validated with the test data that is logged in a second flight test. The UAV dynamics has also been identified by the SSEST method, which is a MIMO identification technique and estimates the state-space model using time or frequency domain data. The results show that the presented method is simple but effective and fast enough to be used for online identification of aerial vehicles and other mechanical systems. The quality of the system identification by the linear neural networks is appropriate and comparable with other MIMO identification techniques and the proposed model is robust against the noise and uncertainties, which can predict all flight parameters.