基于小波阈值的浮动车数据消噪算法

Denoising Algorithm of Express Way Floating Car Data Based on Wavelet Threshold

研究通过小波函数选取等策略来构造一种适合于浮动车原始数据去噪的小波阈值去噪算法.数据去噪是浮动车数据进行交通信息研究的基础性工作,小波分析对于掺杂噪声信号的数据去噪有不可比拟的优势,本文通过构造新的阈值及阈值函数进而以信噪比、均方误差为指导对浮动车数据的去噪结果进行分析来确定小波基函数及小波分解层数,以此研究出针对于浮动车数据的小波阈值去噪的有效算法.通过本文构造的小波阈值算法使得浮动车数据去噪前后与遥感微波检测器(RTMS)数据的相关性提高10%以上,能够有效去除浮动车数据中的噪声.

Data denoising is a foundation work for extracting traffic information from floating car data. Wavelet analysis method shows excellent advantages in data denoising of doping noise signals. In this paper, a wavelet threshold denoising algorithm was proposed to suit data denoising of the original floating car data. According to noise-signal ratio and mean square deviation, the data denoising effect of floating car data was analyzed based on the constructed new threshold value and threshold function to determine the wavelet basis function as well as decomposition level of wavelet, so as to develop an effective algorithm for the wavelet threshold denoising of the floating car data. Results show that, the constructed wavelet threshold algorithm can improve the correlation of denoised floating car data to RTMS data by more than 10%. It means the floating car data can be effectively denoised.