Identifying weak joints in jointed concrete pavements from TSD measurements by basis pursuit

Abstract

Weak joints are the source of most performance problems in jointed concrete pavements (JPCP) and composite pavements created by overlaying a JPCP. Until recently, it has been prohibitively time-consuming to evaluate the joints in a road network, and the evaluation of joint load transfer efficiency (LTE) has been restricted to specific project-level applications. This is changing with the advent of the Traffic Speed Deflectometer (TSD), a device that can collect structural information data at a 1 m resolution while moving at traffic speed. The 1 m resolution is adequate to capture the response of the joints to the applied load, but it results in a higher noise level compared to data collected at lower resolutions (e.g. at the typical 10 m). In this paper, we propose the use of Basis Pursuit (BP) to extract meaningful information about the joints’ condition from the (noisy) TSD measurements. Weak joints are modeled as spikes (Dirac basis) in the measurements while the remaining features in the measurements are modeled by a wavelet basis. Combining the two bases (Dirac basis and wavelet basis) results in multiple possible representations of the collected measurements: we essentially have twice the number of unknowns than the number of equations to determine these unknowns. BP seeks a representation with a small number of elements from the two bases. Because weak joints are best represented by spikes, BP denoising results in mostly selecting the spikes at the weak joints location. These identified spikes provide a list of weak joints that should be further investigated (e.g. by performing LTE testing) or should receive priority in repair. We present examples of BP using simulated data and actual TSD collected data.