Research On Multi-sensor Information Fusion Based Train Location Method

Train location subsystem plays a very important role in Automatic Train Control System. Its accuracy and reliability are important factors for headway in Train Control System. It will affect the efficiency of transport systems. The sensors and the mechanism, which are used in train location subsystem, are related with compatibility of Train Control System and life cycle costs. So it has great significance to study the train location method deeply.This paper introduces the evolution of Train Control System and train location method. According to research on train location method all over the world, it is shown that multi-sensor information fusion based train location method is a research direction in future. In this paper, the structure of multi-sensor information fusion based train location method is studied. An experiment platform of multi-sensor information fusion based train location method is set up. According to the research on the problems in existhing methods, presents an odometer-core multi-sensor information fusion based train location method.In this paper, accroding to research on balance between slide and slip detection and compensation, a novel odometer location error model is proposed. It is founded that error of location is related with threshold of slide & slip detection, in given error conditions. Too high and low threshold would increase error of location. Odometer based location and slide & slip detection based on information fusion with odometer and accelerometer are studied. The effect of accelerometer error to slide & slip detection is analysed. The Effectiveness of slide & slip detection based on information fusion with odometer and accelerometer has been validated by experiment. A novel location method based on the track curvature match is presented. In this method, track curvature is obtained by information fusion with gyrometer & odometer. This method doesn't use high grade gyrometer, and is not sensitive to the gyro drift error and train speed. The Effectiveness of this location method has been validated by experiment. Based on the analysis of the theory of terrain-aided navigation, unity between terrain-aided navigation and balise is accomplished in match location theory. In order to study the trustiness of location method based on the track curvature match, the track curvature entropy is defiend, which is similar with the terrain elevation entropy. The trustness of location method based on the track curvature match is studied by analysis of similarity, and signal-to-noise ratio, and track curvature entropy, and train speed. Experiments results show that the odometer-core multi-sensor information fusion based train location method is effective. It has the alility to detect weak slide & slipe in order to reduce error. This method can measure track curvature, and obtain train location through matching with database of track curvature.