Neural Network Technology And Application Research For Flush Air-data System

Flush air-data system(FADS) utilizes the surface pressure distribution in the vehicle to measure the air-data indirectly. It meets many challenges such as stealth, hypersonic and high angle of attack flight. In this paper, several important issues in the FADS system aiming at the flush air-data system of dynamics model are investigated, including Neural network technology, the influence on prediction accuracy about the number of the pressure taps and the layout of the pressure taps and one or more fault point detection.The CFD technology, by calculating the Reynolds N-S equation, and S-A equation, receives the pressure in different state. Then regard the pressure as the input samples and checking calculation, the air-data calculation model is established by using neural network. The predicted value about three, four and five pressure taps are comparative analyzed. It is concluded that the accuracy of the predicted value of the five pressure taps is higher than that of the four pressure taps, and the accuracy of the predicted value of the three pressure taps is lower than that of the four pressure taps. Specially, taking three points in a straight line can make the predictive value dissatisfy computing requirements. And then, the four pressure taps about atmospheric data are used in aircraft, in order to check the prediction accuracy about the four pressure taps by using neural work.The solution method of the fault point is analyzed. In this paper, two solution methods of the fault point are used, one is using the minimum variance to cluster, and the other is using competitive neural network to cluster. By comparison, the first method is suit smaller number of prediction of pressure points simulated, and the second method is suit larger number of prediction of pressure points simulated. Finally, based on the prediction neural network of three pressure taps like a right angled isosceles triangle, it’s analyzed the relation between the number of pressure taps and the maximum number of the fault point.