A Hybrid LWNN-Based Stochastic Noise Eliminating Method for Fiber Optic Gyro
Dang Shu-wen, Kangle Wang, Hongwei Han, Cheng Peng-zhan · 2016
The output of fiber optic gyroscope (FOG) involves Gaussian white noise and fractional noise which is difficult to be eliminated by traditional methods because of the non-stationary characteristics. Wavelet neural network (WNN) is a novel nonlinear and non-stationary signal processing method, which is exploited in signal denoising. To expedite the computing efficiency and improve accuracy, lifting wavelet transform (LWT) technology is introduced into the WNN method, so that a type of hybrid LWNN-based model is proposed and applied for FOG drift denoising. Experimental results of real drift data show that the proposed model is more feasible and effective in drift denoising compared to the single WT or WNN methods.