No abstract

Car-following behavior is a vital traffic phenomenon in the process of vehicle driving. For modeling the car-following behavior, it is crucial to capture the reaction delay for balancing with safety and comfort, but it is generally ignored in existing works. This work proposes a car-following model based on attention-based ensemble learning to automatically capture the reaction delay from driving data and better depict the traffic flow characteristics. The model integrates a data-driven model and a theory-driven model, and a weight computation method is proposed to combine the advantage of these two different models. In detail, an encoder-decoder model and attention mechanism are employed to capture the reaction delay from driving data. Extensive experiments show that the proposed model could balance safety with comfort and help avoid unsafe driving behavior.

Identifying traffic congestion accurately is crucial for improving the expressway service level. Because the distributions of microscopic traffic quantities are highly sensitive to slight changes, the traffic congestion measurement is affected by many factors. As an essential part of the expressway, service areas should be considered when measuring the traffic state. Although existing studies pay increasing attention to service areas, the impact caused by service areas is hard to measure for evaluating traffic congestion events. By merging ETC transaction datasets and service area entrance data, this work proposes a traffic congestion measurement with the influence of expressway service areas. In this model, the traffic congestion with the influence of service areas is corrected by three modules: 1) the pause rate prediction module; 2) the fitting module for the relationship between effect and pause rate; 3) the measurement module with correction terms. Extensive experiments were conducted on the real dataset of the Fujian Expressway, and the results show that the proposed method can be applied to measure the effect caused by service areas in the absence of service area entry data. The model can also provide references for other traffic indicator measurements under the effect of the service area.

No abstract

The development of the worldwide high-speed rail network is expanding at a rapid pace, imposing great challenges on the operation safety. Recent advances in wireless communications and information technology can integrate the Internet of Things and cloud computing to form a real-time monitoring platform of high-speed trains. To realize this system, a sustainable power source is indispensable. In this case, an ideal solution is to deploy a vibration-based energy harvester instead of batteries for the electrical supply of wireless sensors/devices, as vibrations induced by rail/wheel contact forces and vehicle dynamics are an abundant energy source. To address this challenge, a multi-stable, broadband and tri-hybrid energy harvesting technique was recently proposed, which can work well under low-frequency, low-amplitude, and time-varying ambient sources. In this work, we will introduce our idea, following the recently proposed energy harvester and the dynamic responses of a train vehicle, to design a self-sustained sensing system on trains. Supported by this self-powered system, accelerometers and microphones deployed on an in-service train (in axle boxes/bogie frames) can measure vibration and noise data directly. The correlation of the vibration and noise data can then be analyzed simultaneously to identify the dynamic behavior (e.g., wheel defects) of a moving train.