摘要
为了尽量减小轨道检测数据中夹杂的粗大噪声干扰对轨道检测结果的影响,利用小波算法对轨道检测数据滤波处理是一种可行方法。分析了db1,db2,db3和db4小波基对轨道检测数据中高频突变脉冲信号的敏感性,选用较为敏感的db1小波基对轨道检测数据进行小波分解。通过3σ准则识别出粗大误差点并加以剔除,对轨道检测数据的高频和低频部分进行小波重构,从而达到较好的轨道检测数据去噪滤波效果。分别采用均方误差值、信噪比和平滑度指标对几种小波的去噪滤波效果进行了分析比较,进一步验证了db1小波能在轨道检测数据的处理中达到较好的去噪滤波效果。实例表明,论文提出的方法对能够敏感地识别轨道检测数据中的噪声信号,有效地剔除检测数据中的粗大误差,达到较为理想的轨道检测数据滤波效果。
In order to reduce the influence of gross noise mixed in rail detection data on rail detection, the wave- let filtering algorithm is a feasible approach for rail detection data process. The sensitivity of the high - frequency mutation pulse signals with dbl, db2, db3 and db4 wavelets was analyzed for rail detection data, and dbl wave- let was chosen for the wavelet decomposition of the rail detection data. The gross error points are identified and removed by 3σ criterion, and the wavelet decomposition high - frequency part and low - frequency part in rail de- tection data are reconstructed to eliminate the gross error and noise. RMSE, SNR and smoothness index were used to compare de -noising and filtering results of several wavelets, which further verify that dbl wavelet has better de - noising and filtering effect on the data process of rail detection. The examples show that the proposed method can sensitively identify the noise signals in rail detection data and eliminate the gross error effectively to achieve an ideal filtering effect for rail detection data.
出处
《铁道科学与工程学报》
CAS
CSCD
北大核心
2013年第5期116-122,共7页
Journal of Railway Science and Engineering
基金
中南大学中央高校基本科研业务专项资金资助(2013zzts215)
铁道部科技开发计划重点课题(2011G021-F)
关键词
去噪
粗大误差
小波变换
3σ准则
数据滤波
de -noising
gross error
wavelet transform
3σ Criterion
data filtering
作者简介
韩晋（1990-），男，安徽铜陵人，硕士研究生，从事载运工具运用工程研究
