尖楔前体飞行器FADS-α的求解精度研究

Research on solving accuracy for FADS-α applied to the vehicle with sharp wedged fore-bodies

对用于尖楔前体飞行器的嵌入式大气数据传感系统(Flush Air Data Sensing System,FADS)的不同攻角求解方案(FADS-α)的精度进行研究.对比分析了某尖楔前体飞行器FADS系统的三种攻角实现方案的精度及可行性:(1)将尖楔前缘作为小钝头处理,采用经典的三点式算法建立FADS系统的攻角求解方法;(2)基于压缩波-压缩波理论及膨胀波-压缩波理论建立了FADS系统的理论模型,并发展了相关的迭代算法验证模型的精度及可靠性,建立了FADS系统的攻角实现方案;(3)利用BP神经网络代替FADS系统空气动力学模型的方法,建立了FADS系统的攻角实现方案.针对尖楔前体飞行器FADS系统的特点,设计了一个具有双隐含层的神经网络模型,并对模型的精度进行了验证.结果表明,3种求解方案精度都能满足实际需求.但是方案(1)工程实现困难,方案(2),(3)建立的针对尖楔前体飞行器的FADS系统的求解方案易于实现,且方案(2)的精度优于方案(3).

Solving accuracy for Angle of Attack based on the Flush Air Data Sensing System （FADS-a） applied to the vehicle with sharp wedged fore-bodies is researched. Three different methods for FADS-ct applied to the vehicle with sharp wedged fore-bodies are compared systemically in this paper. （1） FADS-α solving algorithm are developed based on the classic triple algorithm by treating sharp wedged fore-bodies as small blunt fore-bodies. （2） Theoretical model for the FADS system based on the compressible-compressible wave theory and compressible-expansible wave theory has been built, and the related FADS-α algorithm has also been developed to verify the reliability and the accuracy of the model. （3） The method of Back-Propagation neural network algorithm replacing theoretical model of the FADS system has also been researched as to the vehicle with sharp wedged fore-bodies in the paper. In connection with the characteristics of the FADS system applied to the vehicles with sharp wedged fore-bodies, neural network architecture with two hidden layers for the FADS-α algorithm is designed and performed. Among these three methods for FADS-α developed as to the vehicle in this paper, the results show that the solving algorithm and the theoretical model are reliable, and the solving accuracy for all these three methods can satisfy the engineering demands, but allowing for the actual realization, method （1） can hardly be realized because of its small sizes in the fore-bodies. As to the method （2） and method （3）, both of them can be easily realized actually. Allowing for the solving accuracy for angle of attack, method （2） has a better accuracy than method （3）.