您所在的位置:首页 > 中心动态 > 信息通知

【信息通知】美国犹他大学土木工程与环境系助理教授杨险峰博士学术讲座

2019-12-7 16:49:07

 

为了进一步营造科学研究范围,开拓国际学术视野,提升研究能力、学术水平和国际交流能力,大连海事大学综合交通运输协同创新中心于2019年12月10日邀请了美国犹他大学土木工程与环境系助理教授杨险峰博士来我校做学术讲座。

讲座嘉宾:杨险峰 美国犹他大学土木工程与环境系助理教授(终身教职轨迹)、曾任职于美国圣地亚哥州立大学

讲座时间:2019年12月10日(周二)上午9点-11点

讲座地点:大连海事大学西山校区学汇楼208

讲座主题:Machine Learning Applications in Real-time Traffic Management and Control

                    机器语言学习在新一代实时交通管理与控制中的深度应用


1.png


专家简介:杨险峰博士现任职于美国犹他大学土木工程与环境系助理教授(终身教职轨迹)。在任职犹他大学前,杨险峰博士还曾任职于美国圣地亚哥州立大学。杨险峰博士于2009年获得清华大学工学学士学位,分别于2011年和2015年获美国马里兰大学(全美排名前20)土木工程系硕士和博士学位(交通工程方向)。杨险峰博士研究方向主要包括:connected automated vehicle环境下的交通运营与控制,下一代智能交通系统,紧急疏散系统设计,交通安全,交通网络建模,和大网络交通仿真。杨险峰博士课题组研究方向已获美国多个机构资助,其中包括美国国家自然基金,美国交通部,犹他交通部和美国公路局,截止至2019年5月共参与主持超过一百五十万美金的项目,发表80余篇学术论文(40篇SCI/SSCI收录),受邀发表超过40次主题演讲。杨险峰博士目前还兼任TRB Traffic Signal System (AHB25) 和Emergency Evacuation (ABR30)学术委员会的核心委员,并作为创始主席建立ABR30青年学者委员会,同时任美国土木工程协会下属期刊Journal of Urban Planning and Development编委和美国国家自然基金和TRB NCHRP项目评委。

Abstract:During the past decades, researchers and engineers have made great efforts to improve traffic-responsive algorithms of real-time traffic management and control. However, many of them rely on complex optimization functions and may not be applicable on large scale problems. More recently, the ability of machine learning (ML) techniques to capture the stochastic characteristics of traffic has been approved and ML-based control models have shown promising potentials in addressing the curse of dimensionality. In this lecture, two types of ML applications in transportation engineering, including freeway traffic state estimation and adaptive traffic signal control,are introduced. The freeway traffic state estimation is based on a physics-guided machine learning approach which creates a new training variable based on the second-order traffic flow model. Grounded on a novel integrated framework, the estimation is performed using three machine learning techniques, i.e., Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Artificial Neural Network (ANN). For adaptive traffic signal control, a Deep Q-learning Learning (DQL) method is employed. More specifically, a Deep Q Network (DQN) is proposed to estimate the value function of signalized intersections and a deep model is investigated to extract features from flawed data and predict the traffic status, which outperforms the artificial data simulators regarding the running time.