Intelligent Pavement Porosity Assessment Using Ground Penetrating Radar and Neural Network Modeling

Traditional asphalt pavement void ratio and moisture content detection methods are mostly destructive and inefficient. To solve this problem, this paper proposes a non-destructive detection method that integrates field testing and numerical simulation and realizes data acquisition and processing based on the MALA ground penetrating radar system. First, the working frequency of MALA GPR was simulated and analyzed with the help of Gprmax software to determine the optimal detection frequency. On this basis, an indoor rutting plate test was carried out to obtain the corresponding moisture content, void ratio and dielectric constant data, and a prediction model was established through linear fitting, and the results were well fitted. Further combined with the BP neural network, 10 parameters such as stone grading, asphalt type, antenna height, frequency, and waveform amplitude at each moment were used as input variables to carry out prediction modeling. After 500 iterations of training, the model’s prediction accuracy for void ratio and moisture content reached 85.4%, which basically meets the needs of engineering applications.