BP网络在嵌入式大气数据传感系统中的应用

研究了BP人工神经网络在嵌入式大气数据传感系统中的应用.提出了以BP网络为基础的动、静压和修正参数的改进算法,并对其应用Matlab软件进行了验证.改进算法首先应用广义逆矩阵简化方程组,然后应用BP网络求解非线性方程.计算结果表明,改进算法在精度、可靠性等方面可以满足系统的设计要求.在实时性上由于改进算法避免了迭代运算,达到同样的精度所需要的计算时间只相当于原有算法的5%,比迭代方法具有更大实时性优势.

Application of Back Propagation in Flush Airdata Sensing System

The application of back propagation in flush airdata sensing system is studied, and an improved algorithm based on back propagation is presented for calculating the impact pressure, static pressure and modifying coefficient. Its validation is given using Matlab. In the improved algorithm, moore-penrose pseudo-inverse is used to simplify the equations. Back propagation is then used to solve the non-linear equations. The results of the validation indicate that the improved algorithm meets the requirements of precision and reliability. As for realtime, because the improved algorithm avoids iteration, it needs only 5% time comparing to the original algorithm to reach the same precision. It has great advantage on real-time.