BP神经网络在多位置捷联寻北系统中的应用

为了精确拟合多位置捷联寻北系统采集的数据的曲线，计算陀螺初始位置和真北方向的夹角，简要介绍了多位置捷联寻北系统的工作原理，推导了寻北测量的数学模型，并分析了影响测量精度的因素，分别采用了最小二乘法和BP神经网络法，对两种方法的拟合精度和最终计算得到的寻北结果进行了比较。实验结果表明：与最小二乘法相比，BP神经网络拟合精度较高，拟合残差和较小，达到0.06位，残差的均方差达到34.45位，在计算相位角时，多次寻北结果的均值基本一致，但均方差明显由于最小二乘法，达到8″。满足寻北系统对数据拟合精度的要求。

Application of BP neural network to multi-position strap-down north seeking system

In order to fit the sin wave of Strap-down north seeking system and calculate the angle between the start position and the real north accurately, the principle of the system are introduced, and the factors that influence the precision are analyzed. Least square method and back-propagation neural network are adopted separately. Experimental results indicate that the back-propagation neural network is more precision than the least square method, the sum of errors is 0.06, the squared error of errors is 34.45, both of the method get the same average of phases in six measurements, the squared error of BP neural network is 8″, it’s more accurate than the result of the least square method. The back-propagation neural network can satisfy the system precision requirements of curve-fit.