Abstract：Current railway positioning systems primarily utilize GYK and wheel encoders for localization. However, there are challenges such as GYK signal drift and cumulative errors in wheel encoders, coupled with increasing demands for higher automation and precision in railway positioning, rendering standalone GYK or wheel pulse-based positioning inadequate. This paper proposes a particle filter algorithm that integrates wheel encoder and GYK signals to reliably estimate positioning information even when individual signal sources are unstable. The algorithm represents the initial position of the train using particle probability density and employs particle filtering to predict motion trends, achieving robust fusion-based localization. By combining particle filtering with data interpolation, this study enables precise train position estimation. Experimental results demonstrate that, under controlled variables and optimized parameters, the algorithm limits localization deviation to (0.02±0.01) km while reducing positioning errors caused by GYK signal drift. Additionally, the algorithm exhibits strong real-time performance and high repeatability, demonstrating practical value for railway applications.
2025, Vol. 26 
