摘要
针对钝头机体用嵌入式大气数据传感(flush air data sensing,FADS)系统的4类攻角求解算法及算法的求解精度进行研究.针对典型的15°钝头体外形,在Mach数Ma=2.04,3.02,5.01,攻角α=-5°~30°,侧滑角β=0°的条件下,首先基于势流理论及修正的Newton流理论建立了钝头机体用FADS系统的理论模型,并给出了典型的测压孔配置方案;然后采用经典三点式及改进三点式算法、基于线性理论的五孔探针算法、基于非线性理论的五点拟合算法、基于神经网络建模的方法及基于压力模型的加权最小二乘迭代算法,分别建立了FADS-α的4类求解算法;最后对钝头机体用FADS-α的算法求解精度进行了系统对比及论证.研究结果表明,三点式算法、改进三点式算法与加权最小二乘迭代算法精度相当,都可以比较准确地预测攻角;神经网络算法精度较好,但算法涉及的经验参数较多,且需要大批量数据集的训练及验证;拟合算法优劣明显,基于线性理论的五孔探针算法精度在小攻角时与上述几种算法精度相当,但随攻角增大(大于10°)精度下降显著;而基于非线性理论的拟合算法精度较好,但拟合过程复杂繁琐.对钝头机体用FADS-α的算法精度而言,三点式算法、改进三点式算法及加权最小二乘迭代算法是较好的计算方法.
Solving algorithm and accuracy for FADS-αapplied to the vehicle with blunt fore-bodies were researched in this paper.For the typical 15°blunt fore-bodies,FADS operating in the flight envelope from Mach number 2.0 to 5.0 andα=-5°~30°,β=0°was considered.Firstly,the theoretical model based on a potential flow model and modified Newtonian flow theory was established,and the pressure port configuration for the FADS system applied to the vehicle with blunt fore-bodies was also determined.Secondly,the classical triple algorithm and modified triple algorithm,the linear five-hole probe algorithm,the non-linear five-port fitting method,the back propagation neural network modeling and the weighted least square algorithm were established respectively for FADS-αsolving algorithm.Finally,the solving accuracy of different methods for FADS-αwas compared systematically.The results show that the solving accuracy of classical triple algorithm,modified triple algorithm and weighted least square algorithm is similar,and all these three different methods can give good prediction for FADS-α.The neural network modeling can also give good prediction,but its algorithm architecture is determined by many empirical parameters.In addition,it also needs large qualities of training and testing data set.For the linear five-hole probe algorithm,its solving accuracy is similar to other methods for the range of small angle of attack.When the angle of attack is larger than 10°,solving accuracy becomes bad.Solving accuracy is good for nonlinear fitting algorithm,but the fitting process is quite complicated.Therefore,the methods based on the classical triple algorithm,the modified triple algorithm and the weighted least square algorithm are good choices for FADS-αapplied to the vehicle with blunt fore-bodies.
作者
王鹏
WANG Peng(China Academy of Aerospace Aerodynamics,Beijing 100074, China)
出处
《气体物理》
2021年第1期52-61,共10页
Physics of Gases
关键词
嵌入式大气数据传感系统
钝头体
算法
攻角
精度
flush air data sensing(FADS)
blunt fore-bodies
algorithm
angle of attack
solving accuracy
