Author: Osama Abu DaoudPublisher:
ISBN:
Category : Gravel roads
Languages : en
Pages : 146
Book Description
Gravel roads form about 90% of the local roads network in Wyoming. These roads are managed and maintained by local agencies. Generally, local agencies have good experience in gravel roads maintenance execution, while their experience is limited in gravel roads management, maintenance planning, and budget allocation. Moreover, the limitation in available resources for these local agencies makes it more challenging to optimize gravel roads maintenance plans. Hence, the Wyoming Technology Transfer Center (WYT2/LTAP) is currently in the process of developing and producing a new holistic Gravel Roads Management System (GRMS). The proposed GRMS aims to provide local agencies with cutting-edge tools to assist them in managing and maintaining gravel roads under their authority. In addition, the developed tools will significantly enhance the budget allocation optimization by adapting advanced understanding of the gravel roads deterioration behavior. The main purpose of this research is to develop prediction tools and models for gravel roads deterioration using machine-learning techniques. This work started by developing an integrated database. Laramie County in Wyoming was chosen for the pilot study. An intensive effort was invested in collecting gravel roads condition data from the field and building the dataset. The adapted data collection procedure employed two methods: course survey, and smartphones data collection. Afterward, the collected data was analyzed using various techniques to fulfill the purpose of this research work. The employed techniques included: regression modeling, Inception V-3 model, image recognition, TensorFlow, binary logistic regression, random parameter, and Artificial Neural Network (ANN). These techniques will enhance the understanding of the gravel roads deterioration behavior, and the maintenance planning process. This enhancement will significantly improve the GRMS. As a result, the overall gravel road condition will be improved and the road user’s satisfaction will be increased. This research work was designed to be a baseline for employing machine learning and artificial intelligence in new GRMS. Adapting such techniques will produce cost-effective tools for both gravel roads data collection and data analysis. Hence, this research work started by developing an image classifier to rate gravel road corrugation condition using smartphones, the developed classifier is considered a low-cost tool with adequate accuracy. In addition, a sophisticated objective model was developed using the ANN techniques to predict the riding quality on gravel roads with eliminating the subjectivity. The deterioration behavior is an essential parameter in developing any GRMS. So, various gravel roads deterioration models were developed using ANN techniques to forecast the International Roughness Index (IRI). The developed IRI models considered the variation in traffic loading levels and land uses. However, some gravel road sections have special features that affect the deterioration behavior. Thus, the effect of horizontal and vertical curves on gravel roads conditions were studied and analyzed. Lastly, one of the main goals of maintaining gravel roads in adequate condition is enhancing the safety on these roads. Therefore, this research study invested an intensive effort in modeling the crash severities on these roads using mixed binary logistic regression.