Power spectrum density (PSD) analysis is commonly used to determine the distribution of profile energy among its constituting wavelengths. However, the road geometrical curve in the profile introduces artifacts in calculation of PSD functions and accounts for the dominating low frequency portion of the power spectrum, which often masks the variations in the amplitudes of higher frequency components which are of interest in pavement smoothness evaluation. To distinguish different frequency features in the profile and improve the PSD estimation of pavement profiles, it is necessary to remove the geometry trend. In the first part of this paper, a number of issues concerning the common practice of profile filtering are discussed and an alternative filtering operation is proposed. Application of wavelet transforms as advanced diagnostic tools is proposed in the second part of the paper. Although PSD analysis provides the wavelength content of profiles, it cannot be used to locate various frequency features. Wavelet transforms, on the other hand, provide a 2D representation of the distance-varying spectrum of the profile in which frequency features (e.g. repeated waves and short-lived surface distress) can be recognized and located. Advantages of wavelet analysis over PSD analysis are illustrated by presenting the results of the analyses performed on a number of profiles, taken from data collected as a part of a New Jersey Department of Transportation (NJDOT) sponsored research project. The results indicate that wavelet analysis can capture both high and low frequency features of the profile, and, consequently, provides a better representation of the profile smoothness characteristics.
Samenvatting