Research On A Map-matching Algorithm For Train Positioning

The low-cost train control system can not only reduce the wayside equipments, but also cut the system operation and maintenance cost. The reliable train positioning based on GPS is the foundation to realize the low-cost train control system. Since the direct GPS positioning information has potential position error and drift, map-matching methods are often adopted to obtain a higher train positioning accuracy. The existing research on map-matching algorithms concentrates much on the highway application. It is of important practical significance by fully using the unique railroad electronic map information to implement real-time positioning and improve the positioning accuracy.The reliable and accurate electronic map is the foundation to improve the train positioning accuracy. Firstly, in view of "the regularization" two-dimensional model of the railroad track, the curve partition modeling is used to model the rail track to construct the electronic map database in the thesis. Secondly, in the data pre-processing part, the redundancy positioning information of the GPS output is reduced when the train is stopping (or the train speed is lower than a threshold value); the interpolation algorithm is adopted to supply extra positioning information needed for further positioning processing when the train is moving. In the map-matching algorithm part, a line matching method based on the sliding-window related to the travel distance (pre-defined according to the practical demand) is designed to meet the real-time requirement. Regarding the matching algorithm accuracy, the higher positioning accuracy is achieved via an error-compensation method. For the train running in the switch section, a matching method based on the area and distance mixed theory is proposed to judge which branch of the track section the train is moving on. In addition, optimal classification plane is obtained by trained with the original switch section data. Then the train running direction is identified by utilizing the pattern classification's method.Finally, data gathered from San-Jiadian station is used to validate the respective algorithms proposed above. The experimental results show that the map-matching algorithm proposed in the thesis can not only enhance the matching accuracy, but also realize the path identification in the switch section effectively.