System identification using flight test data based on time-domain method is an accurate way of getting a reliable mathematical aircraft model. This thesis provides a system identification procedure on a canard configured fixed-wing aircraft Long-EZ, which is the early and critical stage of providing accurate aircraft models for designing an effective autopilot in the future. 
Flight test designed for Long-EZ aircraft has been carried out by International Test Pilot School (ITPS Canada Ltd). The real flight test data recorded from the testbed has been utilized for the identification and verification of a linear transfer function model, a nonlinear neural network model, and a block-oriented model consisting of linear and nonlinear parts. The linear transfer function structure has been determined with aircraft’s physical dynamics, and the model parameters have been identified using MATLAB System Identification toolbox. The nonlinearity of the aircraft dynamics has been treated with a Multilayer Perceptron (MLP) neural network structure, which has been developed with a set of Python codes. Flight data has been utilized to train this MLP structure. 
The results demonstrate different predicting capabilities of the developed linear, nonlinear, and combined linear and nonlinear structure, which is also known as the neural network Wiener model. The developed Wiener model in general shows satisfactory predicting capability for the testbed Long-EZ aircraft.