Monday, March 12, 2012
Fang Presents Two Research Papers at TRB Meeting
Co-authored with graduate student Patrick Creary and Alireza Jamalipour, P.E., of the Connecticut Department of Transportation, the paper entitled "Development and Practical Application of Artificial Neural Networks in Bridge Level Condition Prediction" developed a revised algorithm. The neural nets algorithm is capable of producing an accurate prediction down to a root mean square error of 10.05% on the best trial. This study considers data mining of about 10,000 inspection records from more than 5,000 bridges in Connecticut during the time span of 2006 to 2009.
The research titled "An Evaluation of the HCM 2010 Operational Analysis Methodology for Interchange Ramp Terminals Using Field Data" was co-authored with colleagues at the University of Florida and is funded by the National Academies. Data was collected at eight sites (two in Connecticut) representing various types of interchanges, with different geometric and traffic characteristics. The research developed a new lane utilization model, modified the effective green time estimation and calibrated queue length prediction. The research findings were incorporated into the new edition of the Highway Capacity Manual 2010, the single most cited traffic engineering guide worldwide.
The TRB Annual Meeting is the premier conference in transportation, attracting more than 10,000 attendees each year. The conference has a research paper acceptance rate of less than 50%.