摘要
轮轨力应变信号在采集过程中,由于噪声干扰的存在,将严重影响所采集数据的准确性。针对轮轨力应变信号中存在的基线漂移和随机白噪声,提出基于小波变换的去噪方法:采用db 6小波基,根据小波多分辨率分析理论,以大尺度分解的逼近分量估计基线漂移成分,从而消除基线漂移;对于随机白噪声则是运用小波阈值去噪法,先根据离散有限序列的自相关函数确定小波分解的最优分解层数,然后采用最小最大阈值以及硬阈值函数,从而实现对白噪声的滤除。仿真与实测数据分析都表明该去噪法能达到比较理想的效果。
The accuracy of the wheel-rail strain signals can be seriously ruined by the disturbance of noise. In this paper, a denoising method based on wavelet transform was proposed for elimination of baseline drift and random white noise. The baseline drift was eliminated by using db6 wavelet bases and the estimation of high-1evel approximation based on wavelet multi-resolution analysis. While the random white noise was eliminated by applying wavelet threshold denoising method. First of all, the optimal decomposition level of the wavelet transformation was determined by applying the self-correlation function of discrete finite sequence. Then, the minimum and maximum thresholds and hard shrinking function were adopted to filter the white noise. The analysis of simulation and the measured data show that this denoising method can achieve ideal effect.
出处
《噪声与振动控制》
CSCD
2016年第1期101-105,共5页
Noise and Vibration Control
基金
国家自然科学基金项目(51478184)
国家自然科学基金项目(51368021)
江西省优势科技创新团队计划项目(20133BCB24007)
关键词
声学
轮轨力
小波变换
去噪
基线漂移
白噪声
acoustics
wheel-rail contact force
wavelet transform
de-noising
baseline shift
white noise
