摘要
设计了基于径向基函数(RBF)神经网络的温度误差补偿方案,并对该方案所采用的标度因数误差模型和偏置误差模型进行研究。根据光纤陀螺的温度误差分布情况设计了标度因数误差和偏置误差联合补偿的方案,将基于多尺度分析的噪声和趋势项分离算法应用于建模数据预处理,以提高建模数据的准确性。建立了RBF神经网络模型,并改进模型的学习方法以防止网络的过拟合。最后,讨论模型输入向量对神经网络规模的影响。温度补偿的结果表明:标度因数误差模型的残差均方(RMS)达到0.73(bit/((°)/s))2,偏置误差模型的RMS达到0.051(bit/((°)/s))2。该建模方法可以消除数字闭环光纤陀螺温度误差,满足中、高精度光纤陀螺实时温度补偿的要求。
A scheme based on Radial Basis Function （RBF） neural networks was designed for temperature error compensation and the scale factor error model and the bias error model were investigated. Based on the temperature error distribution of Fiber Optic Gyroscope （FOG）, a scheme combined scale factor error compensation with bias error compensation was designed for temperature error compensation. A separate algorithm based on multiscale analysis was used in the preprocess of modeling data for higher modeling accuracy. Then, the two RBF neural network models were developed and their learning algorithms were improved to avoid over-fitting. Finally, the effects of the models＇ input vectors on the models＇ scale were discussed as well. The simulation results indicate that Residual Mean Square （RMS） of the scale factor error model is 0.73 （bit/（（°）/s））^2 and the RMS of the bias error model is 0. 051 （bit/（（°）/s））^2. The error models can satisfy the requirements of real-time temperature compensation for mid and high precision FOGs.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2008年第2期235-240,共6页
Optics and Precision Engineering
基金
国家863高技术研究发展计划资助项目(No.2006AA801107)
关键词
光纤陀螺
神经网络
温度误差
误差模型
误差补偿
fiber optic gyroscope
neural network
temperature error
error model
error compensation
作者简介
金靖（1975-），男，内蒙古包头人，讲师，在读博士，主要从事光纤传感器、微弱信号检测等方面的研究。E-mail：jiniing@buaa．edu．cn
张春熹（1968-），男，湖南人，博士，教授，博士生导师，主要从事光纤传感、信号检测等方面的研究。
