Train Localization Environmental Scenario Identification Using Features Extracted from Historical Data

Abstract

The application of Global Navigation Satellite System (GNSS) on the railway greatly reduces the cost on train localization. However, the railway environment is complex and changes with the train movement, buildings, trees, railroad cuts and mountains will block and reflect the GNSS signals, which will bring errors to the GNSS-based train position estimation. This paper proposes a railway scenario identification method based on historical GNSS receiver observation data to identify scenarios along the railway. Firstly, a railway environment scenario parameter model library is established according to Feature of Sky Occlusion (FSO) of typical scenarios, apply historical GNSS observation data along the railway to establish the FSO models of scenario segments, and generate FSO feature sequences. The dynamic time warping algorithm (DTW) is used to match the FSO parameter model of the scenario segment with the FSO model library. This paper collected data from field experiments at Beijing Sanjiadian station to verify the algorithm. The scenario identification results showed that the scenario identification method based on DTW can effectively identify the railway scenarios.