Development Of Data Processing Module In Performance Prediction System For Pavements

The pavement performance prediction model is a kind of tool which can helppavement managers to understand roads conditions. It can predict some of the pavementperformance change. Pavement managers could get better understanding for pavementcondition. So it could be easier for pavement managers to optimize the process ofmanagement in the respects of costs, time, materials, manpower, and so on.This article sets up the model by regression based on more than10million recordsaccumulated by Mississippi Department of Transportation (MDOT) in the past nearly20years.70percent of data is used for regression modeling, while30percent of data isused for model checking. Several important parameters of maintenance model arepicked out as targets in the process of modeling. A set of power form models forpavement performance are set up by nonlinear regression. The new models are dividedinto five families: the Original Flexible Pavement (FLEX), the Overlaid FlexiblePavement (OFLEX), the Composite Pavement (COMP), the Jointed Concrete Pavement(JCP), and the Continuously Reinforced Concrete Pavement (CRCP).25functions areset up including8targets: percentage of alligator cracking for medium and high severity(WCAMH), percentage of other cracking (OC), percentage of other cracking formedium and high severity (OCMH), International Roughness Index (IRI), averagerutting depth (RUT), Pavement Condition Rating (PCR), percentage of spalling (SP),and percentage of cracking (AC). Following work is effectiveness and correlation testfor the new models with check data.After that, the comparative analysis of prediction results between the new andoriginal models is done. The results show that the original prediction models ofpavement performance concentrate more on the evaluation of pavement performance,but are week at the judgment of pavement performance on future trends. Theenvironment today is quite different from many years ago when the model was set up.The prediction accuracy of the original model for the pavement performance hasdeclined. While the models set up in the article summaries the previous experiences.The models can predict the trends of pavement performance in the future, after theinfluence of road age is considered in the models. The prediction precision is improvedby considering more factors effecting on pavement performance. Following that, theconversion factors of another model for pavement performance prediction, the DensityModel, are modified with the new models by Bayesian Method. The prediction accurate of the Density Model is increased. At last, the article introduces the effectiveness of thenew model by an actual case.This set of prediction models for pavement performance connects well to therehabilitation and maintenance model for pavement in the original pavementmanagement system (PMS), in the condition of keeping all other parts of the originalPMS without influencing the normal running. The prediction accuracy of the PMS forpavement performance is increased, analyzing from the checking results and practicaleffect. This set of prediction models for pavement condition can match well with therequirements of pavement managers for understanding of pavement condition. This newset of prediction models for pavement performance can improve and enrich theevaluation method of PMS for pavement performance. More reference conditions canbe provided to pavement managers. The contribution for better management of roads isobvious.