Development of rutting forecasting models for distinct asphalt pavement structures in RIOH testing track using different approaches

Rutting is one of the main exacerbation phenomena of the asphalt pavements, which seriously affects road safety and the quality of service. The predicted models are indispensable for preventing and controlling the damage caused by the exacerbation of pavement management systems. This paper is centered on developing more appropriate and exact rutting forecasting models for estimating the rutting performance of distinct asphalt pavement structures on the basis of the field data set of the Research Institute of Highway (RIOH) testing track. First of all, the critical influential factors giving rise to the rutting evolution are identified through analyzing the field data set of the RIOH testing track. After that, the framework of the rutting forecasting models based on the artificial neural networks (ANNs) coupled with genetic algorithm (GA) (GA-ANNs-based rutting forecasting models), the frameworks of the experiential and mechanics-based experiential rutting forecasting models are established according to the implicit relationship between the rutting and the key influential factors. Then, the unknown parameters of the experiential and mechanics-based experiential rutting forecasting models for different asphalt pavement structures are designed by the multivariate nonlinear optimization algorithm. Finally, the applicability of the GA-ANNs-based rutting forecasting models, the experiential and mechanics-based experiential rutting forecasting models are discussed and illustrated through analyzing their forecasting effects. The result illustrates that the proposed rutting forecasting models have better applicability for different asphalt pavement structures (i.e., the semi-rigid base pavement structure, the rigid base pavement structure, the flexible base pavement structure, and the full depth AC pavement structure).