Developing Enhanced Performance Curves of ITD Asphalt Pavements by Mining the Historical Data

Presentation

Developing Enhanced Performance Curves of ITD Asphalt Pavements by Mining the Historical Data

This presentation introduces work of a recently finished project aiming to develop reliable and realistic and enhanced performance curves for Idaho Transportation Department (ITD) asphalt pavements by mining the historical data. To this end, the project reviewed currently applied predictive models in terms of their forms, applications, advantages and limitations. Besides, a practitioner survey was conducted on the insights and experiences of users on the existing models for the basic qualities of a predictive model should have to be applied in practice. According to characteristics of historical data collected by ITD, machine learning (ML) models of different types such as neural networks (NN) and gene expression programming (GEP) were utilized and compared with traditional models such as mechanistic-empirical models and piecewise linear regression models. In addition to model accuracy, this project paid attention to basic applicability of predictive models, statistical methods were utilized to check the stability, robustness, sensitivity, etc. of constructed models before application.

Speaker

Yong Deng, PhD, Research Assistant Professor
Washington State University

Dr. Yong Deng is currently a research assistant professor of the Department of Civil and Environmental Engineering at Washington State University (WSU) and a researcher at the National Center for Transportation Infrastructure Preservation and Life-Extension (TriDurLE). He obtained his Ph.D. and M.S. degrees in Civil Engineering from Texas A&M University in 2017 and 2020. His current research focus and interests are finite element (FE) model updating of pavement materials and structures, data-driven models for pavement performance evaluation and prediction, applications of artificial intelligence algorithms, etc.

Register Here