Research On Fault Detection And Isolation Of Integrated Train Positioning System

Train Positioning System is the basic of Train Control System to ensure the train works reliably. Train's position is measured by Multi-sensor based Integrated Train Positioning System, which utilizes many positioning sensors, fuses data from each sensor and outputs position information. System reliability and fault tolerance ability are improved. Redundant information is utilized to detect fault and the other subsystems are reconstructed to assure train positioning reliably. The essential technology is Fault Detection and Isolation (FDI).Firstly some FDI algorithms were analyzed and compared. Wavelet Transform based FDI algorithm was addressed. Sensor fault detection utilizing singularity of fault signal was discussed. Fault normally acts as high frequency signal or signal's singular point. Local maxima of wavelet transform, which embodies signal's singularity, could detect high frequency signals. After wavelet transform, local maxima was processed and detailed information was extracted. Fault was judged according to fuzzy logic based strategy. Fault sensor was isolated and system was reconstructed according to isolation strategy. So position information was supplied continuously.Integrated train positioning system platform was designed to verify the FDI algorithm. Devices such as sensors and on-board computer were integrated on that platform. Fault simulation of each sensor was processed on the platform via hardware-in-loop (HIL) simulation. The simulation method was that adding some fault signals on the test data from GPS/IMU/ODO integrated system. Sensor's fault was detected via wavelet transform based fault detection algorithm. The effect of fault detection on system positioning was verified. So the simulation has practical value.As a conclusion, sensors' fault was detected by FDI algorithms successfully. False alarm rate and missed detection rate were declined. The fault was positioned accurately. Fault sensor was isolated and system was dynamically reconstructed. The positioning precision and reliability were improved. The effect of the positioning system was verified.