摘要
针对半球谐振陀螺输出信号噪声的特点,提出了一种处理该噪声的前向线性预测(FLP)滤波与小波变换相结合的滤波方法,运用Allan方差法对滤波结果进行了分析。该方法以FLP滤波作为前段滤波器,采用DB4小波函数的强制阈值小波变换作为第二级滤波器。为了改进该级联滤波算法的不足,提出了基于神经网络的滤波融合算法。仿真结果表明,该滤波融合算法保留了级联滤波算法的优势,同时最大程度的保留了信息的完整度,滤波精度得到极大提高。
Aiming at the noise of HRG signal,a cascaded filter method that combines the forward linear prediction（FLP） filter and wavelet transform was proposed,and the results of cascaded filter were analyzed by Allan variance.The method uses the FLP filter as a preceding filter and uses the DB4 wavelet function mandatory threshold wavelet transform as second filter.In order to improve the cascaded filter algorithm,a fusion algorithm based on neural network was proposed.Simulation results showed that the filter fusion algorithm retains the advantages of cascaded filtering algorithm,while the maximum reservation information of integrity,filtering accuracy is greatly enhanced.
出处
《现代科学仪器》
2013年第2期85-88,共4页
Modern Scientific Instruments
关键词
半球谐振陀螺
FLP滤波
小波变换
神经网络
Hemispherical resonator Gyro
Forward Linear Prediction Filter
Wavelet Transform
Neural Network
