Prediction of aircraft aerodynamic coefficient based on data-driven method

Abstract

To identify the aerodynamic coefficient of aircraft that cannot be directly measured in actual flight, in this paper, the problem of aerodynamic coefficient identification has been solved with the rigid body three-degree-of-freedom model using aircraft dynamics model based approaches. And based on the data-driven method to predict the aerodynamic coefficients, based on radial basis function neural network (RBFNN) with particle swarm optimization (PSO) has been used to achieve the task. Taking the external disturbance and the noise of flight data into account, a section of flight data of the XXX aircraft is preprocessed. The aerodynamic coefficients are identified based on the pre-processed flight data, and the prediction results of the aerodynamic coefficients of the above methods are analyzed, which proves the effectiveness of using the data-driven method to predict the aerodynamic coefficients.