In new Automatic Train Control (ATC) and Automatic Train Protection (ATP) systems, the onboard navigation system plays a vital role. Applying multisensor fusion algorithms to vehicle navigation has made it possible to build inexpensive and accurate positioning systems which will satisfy the requirements of railway systems. Since the reliability of the navigation system is of as great a concern as its accuracy, it requires the design of a navigation system in which the detection and isolation of faults is possible. In this paper, the authors propose a multifilter architecture for a multisensor navigation system which can be used for onboard train navigation, explaining the methods for fault location and isolation of the system. Two different fault detection techniques which can be used besides the Kalman filter algorithms are explained and their performance is compared in the simulation work described at the end of the paper.
展开▼