Aircraft Unsteady Aerodynamic Hybrid Modeling Based on State-Space Representation and Neural Network

Abstract

This paper proposes a hybrid model which combines state-space representation and back-propagation neural network to describe the aircraft unsteady aerodynamic characteristics. Firstly, the state-space model is analysed and evaluated using wind-tunnel experimental data. Subsequently, back-propagation neural network is introduced and combined with state-space representation to form a hybrid model. In this hybrid model, the separation point model in state-space representation is reserved to describe the time delay of the unsteady aerodynamic responses, while the conventional polynomial model is replaced by back-propagation neural network to improve accuracy and universality. Finally, lift coefficient and pitch moment coefficient data from the wind-tunnel experiments are used to estimate the hybrid model. With high similarity to the wind-tunnel data, the hybrid model presented in this paper is proved to be accurate and effective for aircraft unsteady aerodynamic modeling.