APISAT 2014 Study on Algorithms of Flush Air Data Sensing System for Hypersonic Vehicle

In this paper, according to the typical hypersonic vehicle,a study of flush air data sensing system using BP neural network algorithm were carried out. The independent research and development of CACFD software for solving the Euler equations, calculation of vehicle head pressure distribution as the neural network training input, the corresponding flow conditions, such as the static pressure, the Ma number, angle of attack and side slip angle as the target samples to train the neural network, set up FADS algorithm based on BP neural network and testing . Studies showed that the FADS algorithm based on neural network technique is robust and has good precision, strong real-time, is a very effective algorithm. Research results had showed that : in the sample number range, the precision of FADS is increasing with the the number of samples increasing ; the average error of FADS algorithm decreases with increase of pressure measuring point of layout combinations; contains large cone angle position measurement points, points than only the small cone angle measuring point combination results average error .To remove vertex pressure point, had little effect on the precision of algorithm, the FADS algorithm generalization performance is very stable with 1% pressure measurement error © 2014 The Authors. Published by Elsevier Ltd. Peer-review under responsibility of Chinese Society of Aeronautics and Astronautics (CSAA).