Train operational state estimation: Technologies, challenges, and future directions

As the demand for intelligent transportation systems grows, train operations are becoming increasingly complex, making state estimation critical for improving railway safety and operational efficiency. Existing state estimation methods face significant challenges, particularly in addressing complex dynamic environments and multi-train formation scenarios. This paper provides a comprehensive review of advancements in train state estimation, focusing on the application of Kalman filters while exploring nonlinear non-Gaussian filters, multi-sensor fusion, deep learning, and other advanced algorithms. Research is categorized into classical model-based methods, data-driven approaches, and hybrid models, summarizing key contributions. Challenges related to real-time performance, robustness, and accuracy, particularly in nonlinear systems, non-Gaussian environments, communication delays, and data loss, are discussed. Future research directions are proposed, emphasizing the need to improve estimation accuracy, address nonlinear dynamics, ensure robustness, enhance inter-train information sharing, mitigate delays and packet loss, and reduce real-time computational burdens.