Study On Application Of Particle Filter In Train Integrated Positioning Of The High-Speed Railway

All subsystems of the train operation control system, such as Automatic Train Pro-tection (ATP), Automatic Train driving (ATO) and Automatic Train Supervision (ATS), usethe train position information, especially in High-Speed Railway, the train positioning systemwith high precision is vitally significant for the safety of the train operation and improvementof the train transportation efficiency.At present the train positioning system in domestic railway mainly includes the speedsensor positioning system, traditional track circuit system and Balise System (BaS). Butowing to wheel wear and other reasons, there are accumulated errors in the speed sensor posi-tioning system; and track circuit and BaS system have the poor real-time performance,because they are all discontinuous positioning system.In the navigation and positioning systems, it spends huge to improve certain positioningsensor precision, and the effect is not obvious, multiple sensor fusion can integrate theadvantages of different types of sensors. In Inertial Navigation System (INS) there areaccumulated errors, and Global Positioning Systems (GPS) is greatly influenced by the envi-ronment. INS/GPS integrated positioning can make up for respective disadvantages and pro-vide high precision positioning information, and then BaS which can provide the absolutelocation information is combined, positioning accuracy will furtherly be improved. In this the-sis the basic principle of the INS system and the GPS system are introduced, and based on theposition-velocity loose combination mode, the mathematical model of errors control is formedin positioning system.In filtering method, at present the filter method is mainly Kalman Filter (KF) method,and Expand Kalman Filter (EKF) method, but the two algorithms are limited when used innonlinear system. In this case, Particle Filter (PF) that is suitable for nonlinear system attractswidespread attention. In this thesis the PF algorithm and its essential issue are introduced, andbased on the former research achievement, introducing the Fuzzy Inference System (FIS), andthe Fuzzy Adaptive Hybrid Annealed Particle Filter (FAHAPF) algorithm is formed. Thensimulation is carried out, and the simulation results show that under FAHAPF algorithm trac-king accuracy and the convergence speed have been improved.In the last chapter based on analyzing the BaS system, monolithic fusion structure andalgorithm are formed, and then simulation is carried out respectively based on the INSindependent positioning, INS/GPS combination positioning and INS/GPS/BaS combination positioning, and the simulation results show that the accuracy of INS/GPS/BaS combinationpositioning is highest. Furtherly under the situation of INS/GPS/BaS, PF algorithm andFAHAPF algorithm are compared with simu lation results. The results show that FAHAPFfiltering algorithm has higher tracking accuracy, so it is proved the effectiveness of the fusionmodel and the fusion algorithm.