摘要
介绍了嵌入式大气数据传感系统及其数学模型和校正过程,研究了该系统的BP网络校正算法;提出了以BP网络为基础的迎角、侧滑角和形压系数的校正算法,并对该算法应用MATLAB软件进行了验证;该校正算法采用BP网络,分别以当地迎角、当地侧滑角和马赫数作为BP网络的输入,以真实迎角、真实侧滑角和形压系数作为输出,通过对BP网络进行训练,从而得到系统的校正算法; 计算结果表明,该算法在精度、可靠性和实时性等方面可以满足系统的设计要求;在精度上,由于BP网络对非线性函数的无限逼近特性,可以用更少的参数实现同样的数据精度,易于实现。
The calibration algorithms of flush airdata sensing system based on BP neural networks is studied. The calibration algorithms for the angle of attack, angle of sideslip, and the shape and compressibility parameter is presented based on BP neural networks and the algorithms is simulated numerically in Matlab 7. 0. The algorithms contain three BP networks. It takes the local angle of attack, local angle of sideslip and the roach number as the input of the BP neural networks and takes the real angle of attack, angle of sideslip, and the shape and compressibility as the output. After the BP neural networks being trained, the calibration of the system can be realized. The results show that these algorithms meet the requirements of the system on precision, reliability and real-time. As for the precision, because of the high nonlinear mapping ability of BP neural network, it needs fewer parameters to reach the same precision. Thus this calibration algorithm is easier to realize.
出处
《计算机测量与控制》
CSCD
2006年第4期541-544,共4页
Computer Measurement &Control
关键词
BP网络
嵌人式大气数据传感系统
校正
函数逼近
back propagation
flush airdata sensing system
calibrationt function approximation
