自适应提升小波神经网络光纤陀螺滤波方法

Adaptive Denoising Method Based on Lifting Wavelet Neural Network of Fiber Optic Gyroscopes

采用传统滤波方法很难有效滤除光纤陀螺输出信号中的随机噪声。提出一种基于提升小波神经网络的自适应阈值选取滤波方法对光纤陀螺的输出信号进行滤波,进而提高光纤陀螺的精度。算法包括小波提升格式转换、提升小波分解、自适应阈值选取及小波神经网络滤波。通过仿真实验将传统小波方法、经验模态分解方法与新方法进行比较,实验结果验证了新方法的有效性。

The output of FOG involves stochastic noise which is difficult to be eliminated by traditional methods. An adaptive denoising method based on lifting wavelet neural network was proposed to improve the precision of fiber optic gyroscopes. The algorithm consists of transformation of lifting format, de-composition based on lifting wavelet, level-dependent threshold selecting and denoising based on wavelet neural network. The de-noising method based on wavelet and empirical mode decomposition were also been investigated to provide a com- parison with the new method. Simulation experiment shows that the ALWNN method outperforms the other two methods.