基于随机误差项的小波阈值去噪算法研究

针对振动环境下陀螺仪输出信号噪声干扰严重的问题,提出了一种用随机误差项改进小波阈值的去噪算法。通过对陀螺仪输出信号进行小波分解,根据频率成分将信号分解为多层;然后,对分解在各层的信号进行随机误差项辨识,进而利用随机误差项系数获取各层的噪声阈值;最后,利用获取的阈值进行小波去噪。改进阈值的提出,旨在解决Donoho全局阈值中因阈值选取过大或过小而产生的噪声误判或噪声残留问题,使噪声去除更彻底。通过实验分析,证明了本算法既能有效去除信号噪声,解决噪声残留的问题;又能保留输出的有效信号,解决噪声误判的问题。

Study on Improved Wavelet Threshold De-noising Algorithm Based on Random Error Term

Aiming at the problem that the noise of the gyro output signal is serious in the vibration environment, an improved wavelet threshold de noising algorithm based on the random error term is proposed. Through the wavelet decomposition of the gyroscope signal, the signal is decomposed into multiple layers according to the frequency components. Then, the random error term is identified for the signals decomposed at each layer; and then the noise threshold of each layer is obtained by using the random error term coefficients. Finally, the obtained threshold is used for wavelet de noising. The proposed threshold is proposed to solve the problem of noise wrong judgment or noise residue due to excessive or too small thresholds selection in the Donoho global threshold, so that the noise removal is more thorough. Through the experimental analysis, it is proved the algorithm can not only effectively remove the signal noise and solve the problem of noise residue, but also keep the effective signal of the output and solve the problem of noise wrong judgment.