Research On Train Location Method Based On Kalman Filter For Multi-sensor Information Fusion

In recent years,the high-speed railway has continuously developed.Therefore,the requirements of the safety and reliability of the train control system are getting higher and higher.The train positioning system is the most important part of the train control system.Therefore,the positioning accuracy of the train positioning system must be more accurate.This design of the overall positioning program divides into two parts.In a multi-sensor portion,based on the traditional train positioning subsystem consisting of odometer,Doppler radar and balise,the accelerometer added to design the sensor information fusion architecture.In the information fusion section,Build a suitable information fusion algorithm model and select the Kalman filter algorithm as the information fusion algorithm In the process of designing the overall plan of the train locating system,analyzed the insufficiency of the traditional train locating system and carried out the improvement and verification.In the multi-sensor part,focuses on the odometer,which is the most important sensor.To deal with the problem of position error caused by wheel diameter wear,carry on the corresponding research First,described the existing calibration program and their respective deficiencies.Secondly,improved the gray prediction model,which combined with a calibration program based on Kalman filter to improve the positioning accuracy.Though the measured data validation and MATLAB simulation of these two aspects prove that the design of the wheel diameter correction program does improve the accuracy of the correction and better adapt to the train-operating scenario.The overall program has some practical value.In the part of information,build a foundation for the algorithm and establish a reasonable train movement model On the one hand,analyzed and explained the theory and the problems of the existing train movement model On the other hand,improved current statistical model and designed the interactive train motion model combined with uniform acceleration model In this way,the program better describes the train's movement status.At the end of this thesis,analysis of the three aspects of the data through MATLAB.It is verified that the interactive train motion model designed in this paper better describes the change of the train's motion state and improve the positioning accuracy.Secondly,the dissertation expounds the discrete Kalman filter and its defects and deficiencies.In addition,the defect that its external noise requirement must be white noise and Colbred noises whitened.Aiming at the problem of divergence of filtering and unknown environment explores the adaptive Kalman filter and designs a simple improvement program.Build a real hardware and software system platform and verify the practicality of the system based on the simulated sand table of the Beijing-Shanghai line.The measured results show that the overall program practical value to meet the actual project needs.Through the actual platform verification and the corresponding software platform data analysis,it is verified that the improved positioning scheme designed in this paper does improve the positioning performance of the train positioning subsystem to a certain extent.