GPS/BDS-based Virtual Balise - Enabling Satellite-based Train Control with a Train-centric Approach

Satellite positioning using both the global and local GNSS (Global Navigation Satellite System) signals is becoming a very attractive technique in next generation cost-efficient railway signaling systems. The rapid development of BDS (BeiDou Navigation Satellite System) is providing more opportunities for a joint use of dual or multiple constellations in autonomous localization of railway trains. To better improve the capability of satellite-based train positioning for safety control of the trains, the concept of Virtual Balise (VB) provides a good choice to fulfil the requirements of novel train control system solutions with a train-centric principle. In this paper, a GPS/BDS-enabled LDS (Location Determination System) architecture for realizing Virtual Balise is investigated. The integration of GNSS with additional sensors (i.e. inertial sensors and odometer) is designed to realize seamless and trustworthy train positioning using a nonlinear estimator. Quality evaluation logics are embedded into the positioning determination process to deal with possible harsh observing conditions and faults. With this basis, a model predictive VB capture solution is proposed to improve conventional VB capture method based on a fixed Capture Interval (CI). Both the spatial and temporal resolution of VB capture are enhanced with the assistance from the additional sensors and track map. To establish an enhanced VB solution at the system level, theoretical design and field tests are carried out in this paper. The contributions mainly focus on the following aspects: 1) The multi-sensor-based train positioning using a nonlinear filter and trackmap database; 2) A CI-free VB capture solution that is compatible of different operational conditions and achieves the balanced capture rate and capturing precision level; 3) A hierarchical quality evaluation method covering multiple layers to ensure effectiveness of VB capture. The proposed solution is tested using field data against conventional fixed-CI- based VB capture method. The results illustrate its advantages and potential in developing novel train control systems in the future.