HHT theory in road roughness in the application of signal

sition, referred to as EMD) to decompose the signal into a series of intrinsic mode function, on this basis, proposed a criterion based on the correlation coefficient matrix of empirical mode decomposition method to remove trend term, using simulated signals and measured road roughness signal practicality of the method. Roughness of the road noise signal removal methods. The traditional method of wavelet thresholding small wavelet basis functions and threshold functions such as parameter selection uncertainty, was proposed based on empirical mode decomposition method of wavelet thresholding. Discrete inverse Fourier transform method to generate pure road roughness signal, adding random noise, respectively, based on empirical mode decomposition using wavelet threshold method and wavelet method of treatment, results showed that the threshold criteria based on fixed (Sqtwolog) the empirical mode decomposition wavelet hard threshold denoising the best. Meanwhile, road roughness and the measured signal analog signals, using different basis functions to verify the wavelet threshold and empirical mode decomposition method of wavelet threshold denoising effect and found that the empirical mode decomposition based on wavelet threshold denoising wavelet transform method can weaken the basis function of the threshold denoising. Using this method for denoising the measured surface data analysis results show that the noise reduction effect is good. To the measured surface data, for example, compared four different power spectrum estimation method, the results show that the average periodogram method of Window to the requirements of both the variance and resolution. Preparation procedures, extraction of surface elevation data, using empirical mode decomposition based on the correlation matrix approach to tendency, finite impulse response filter design, filtering the data; based on empirical mode decomposition method of wavelet hard threshold denoising , and then after the data preprocessing power spectrum estimation, enabling processing of raw data throughout the process. 

Key words: power spectral density, HHT, tendency, denoising, EMD