Texture Measurement And Skid Number Prediction Using Laser Data Acquisition, Digital Signal Processing, And Neural Networks

Real-time estimation of the skid number of pavement is difficult. Traditional methods of volumetric measurement are cumbersome and time consuming. It is desired to enable prediction of the skid number of pavement using non-contact means, and to doso using a method which provides a reasonable estimate of the pavements skid number. This research used laser data acquisition of macro-texture, Digital Signal Processing and Neural Networks to estimate the skid number of pavement to a reasonable degree. The research used Digital Signal Processing to identify potentially bad data sets, and a Neural Network model for predicting skid number on the refined data from the DSP. The method enabled relating a statistical index to the texture characteristics of pavements. The model is based on surface roughness characteristics of pavements as measured by a laser based measurement system, and has the potential to be adapted to a real-time measurement system.