Analysis of pavement condition index in Gansu Province based on a Bayesian hierarchical model
A statistical model was established for the natural decay pattern of one of the key performance indicators of Gansu Province's ordinary national road pavements,i.e.the pavement condition index(PCI).Considering the entirely different climate conditions that exist across different regions in Gansu,this study clustered various sections of roads based on environmental data such as rainfall and sunshine dura-tion.Using annual traffic equivalent data(traffic volume for the entire year)and PCI data,a Bayesian hier-archical model was constructed via Gibbs sampling for Bayesian inference.Experimental results demon-strated that,under varying evaluation criteria,the Bayesian hierarchical model outperformed non-hierar-chical models in overall performance on the test set.Through model analysis,the following findings were discovered:annual traffic volume had no distinguishable impact on PCI for different climatic zones along national highways in Gansu Province;road sections with larger initial PCI values tended to have a higher PCI value after a year of decay,and the initial PCI value also served to mitigate the rate of PCI decay.The influence of the initial PCI value varied from region to region,with the initial PCI value having the most significant mitigating effect on future PCI decay for national highway sections in the Hexi Corridor,followed by the south-eastern Gansu area,and weakest in the Lanzhou region and high-latitude areas of eastern Gansu.
pavement condition indexannual average traffic equivalentBayesian hierarchical modelGibbs sampling
万方数据
维普
