Evaluating road surface conditions using tire generated noise

Abstract

Classifications of road conditions are crucial because officials prioritize road maintenance decisions based on them. Pavement condition index (PCI) surveys are performed manually and used by many cities in the U.S. to evaluate road surface conditions. In this research, a more efficient method is used to detect road surface conditions. This method applies a probabilistic analysis to acoustic pressure data collected from a vehicle-mounted microphone. The data is collected while the driving and processed in real time. Acoustic pressure data contains information on road surface conditions because acoustic pressures change when the tire impacts different road surfaces. This change is audible to human ears, for example, a driver transitions from a normal road to a bridge. The acoustic pressure data used in this research was collected from roads with known PCI values that are used as a reference. To reveal the dominant common features and neglect trivial differences within a certain length of road, a probabilistic method is used to evaluate road surface conditions. This approach uses the Weibull probability density function (PDF) to evaluate road surface conditions. This distribution was chosen because it is closest to the actual PDF among other distributions such as the normal distribution and lognormal distribution. A key finding of this paper is that the Weibull PDF shows the largest change between roads with different PCI values. Another finding is that the Weibull pdf changes when the van hits road defects such as cracks and patches.