Aerodynamic derivatives identification for ground vehicles in crosswind using neural network and PCA

Abstract

Principal component analysis (PCA) is employed in this study to reduce the size of the neural network input node. Neural network is used to identify the ground vehicle aerodynamic derivatives based on a recorded simple harmonic motion of a ground vehicle model. The study involves the identification using neural network with and without the input optimisation by PCA. Both studies are compared with the identification results from a conventional method, and it is shown that the neural network can approximate functions based on principal components extracted as well as a full-size neural network can.