基于小波去噪与 DRNN 的光纤陀螺随机误差建模研究

摘要

For the question of modeling the gyro random error which has complex feature , this paper uses the wavelet threshold denoising with good time-domain and frequency-domain localization properties and the diagonal neural network ( DRNN) with local internal recursive nature to study the model .First, there are some theories about wavelet and DRNN .Then, this paper use the wavelet threshold denoising and DRNN in measure data to model the error .Simulation results show that based on the action of the wavelet threshold denoising which can reduce the high frequency noise of gyro random error , DRNN can model the low frequency noise effectively .There is anticipant result by combine with the two theories .%针对具有复杂特性的光纤陀螺随机误差的建模问题，采用具有良好时域和频域局部化性质的小波阈值去噪和具有局部内递归特性的对角神经网络（ DRNN）进行降噪建模研究。首先给出小波阈值去噪和DRNN的相关理论；然后，组合利用小波阈值去噪和DRNN对实测数据进行建模研究。仿真结果表明，在利用小波阈值去噪去除陀螺随机误差高频噪声的基础上， DRNN能够有效地对低频噪声建模，组合运用这两种理论方法取得了理想的建模效果。