AI-Enabled Intelligent Vibration Sensor for Active Highway-Rail Grade Crossings
Overview
Highway-Rail Grade Crossings (HRGCs) have many safety challenges due to the potential collisions between oncoming trains and road users, including vehicles, bicycles, and pedestrians. The implementation of Positive Train Control (PTC) technology is regarded as a promising solution to reduce rail accidents. However, its effectiveness is hindered at HRGCs, especially in rural regions, where radio and GPS communication can be unreliable or lost (dark territory). To mitigate the risk of catastrophe, it is imperative for road users to detect trains early and maintain a safe distance allowing sufficient reaction time. This can be challenging especially in rural regions where it is difficult to supply enough power for advanced and heavy train detection sensors. This research seeks to develop an AI-enabled vibration sensing system capable of identifying and tracking approaching trains from a considerable distance upstream. This advancement enables road users to preemptively initiate responsive actions within a secure timeframe. The system’s design is especially tailored for deployment in remote areas where there is limited access to electric power sources required for conventional vibration sensors.
Research Objectives
This research project in summary aims to achieve the following objectives:
- Studying a highly sensitive vibration system capable of detecting train-induced vibration of the track with ballast ranging from micro to macro levels.
- Developing a physics-informed machine learning model capable of detecting the dynamic characteristics of approaching trains from a significant distance upstream (more than few miles). The data used to train the ML model will include both time-series signals and features in time-frequency domain for a wide range of train’s speed, weight, and effective length.
- Developing a model to predict both the optimum reaction time of the road users and the closure time at the HRGC to ensure smooth and efficient traffic flow.
Personnel
Principal Investigators:
Mohsen Amjadian, Ph.D. PI
Constantine Tarahnew, Ph.D., Co-PI
Deliverables
Coming Soon
Datasets
Coming Soon
Details
Coming Soon