On the Use of Axle-Box Acceleration Data for Rail Vehicle Positioning

Accurate and reliable rail vehicle position and motion information is crucial for many relevant applications in the rail domain. Important examples include the optimized and automated operation of vehicles and the infrastructure, and the monitoring and management of rail assets using in-service data. The article explores how vibration data collected on rail vehicles can help solving the positioning problem. Specifically axle-box acceleration (ABA) sensors are investigated as rich information sources that can complement onboard sensors and systems such as inertial measurement units (IMU) and receivers for global navigation satellite systems (GNSS). ABA data exhibit an interesting duality because their use for track and vehicle condition monitoring requires knowledge about the vehicle position and/or velocity. With intelligent processing schemes, however, they can serve as position and velocity sensors. Methodologies for finding the driven path and on-track position in a track network based on ABA pattern recognition are presented. Furthermore, speed estimation using ABA data is performed using frequency-domain signal processing and neural networks (NN). Results are demonstrated on actual data collected on an in-service rail vehicle.