Adaptive false discovery rate for wavelet denoising of pavement continuous deflection measurements

Abstract

The traffic speed deflectometer (TSD) is a device used to evaluate the pavement’s structural condition. Measurements obtained from the TSD are affected by noise, which can make it hard to interpret test results. The main objective of this paper is to develop a denoising methodology to use with TSD measurements and improve pavement structural evaluation. The denoising methodology comprises a computational algorithm to identify significant features in a high-dimensional vector of observations containing white Gaussian noise. The algorithm minimizes the classification error of features in the wavelet transform domain by adaptively selecting the level at which to control the false discovery rate. When tested in a simulation study, the results of the proposed algorithm compared favorably with other state-of-the-art methods. The proposed methodology was then successfully used with TSD measurements to identify possible weak joints in a jointed concrete pavement overlaid with an asphalt layer and to calculate asphalt layer modulus values. Repeated measurements were used to validate that the denoised measurements more accurately represent the structural condition than raw measurements.