Hybrid Fault Detection System in the Presence of Transient Faults

This research considers the fault detection method to detect abnormalities in the presence of transient faults using Kalman filter and Adaptive transient fault model (A-TFM). Kalman based fault detection system is used to determine whether the abnormality occurs or not by comparing the estimated value from Kalman filter and the measured vale from sensors. If the difference (i.e., residual) between two values is higher than a predetermined threshold, the system makes a fault alarm. However, the transient fault (e.g., GPS signal in a tunnel) could happen in a normal operation, and when this fault is considered as abnormalities, it makes a false alarm. To address this problem, we propose the hybrid fault detection method using Kalman filter and A-TFM that detects abnormalities in presence of transient faults and provides an adaptive threshold. In addition, we validate the effectiveness of our proposed method with real measurement data obtained from an unmanned ground vehicle called Jackal.



  • Minsu Jo (Minsu-Jo@dgist.ac.kr)
  • Soohyeon Kwon (shkwon@dgist.ac.kr)
  • Youngmi Baek (ymbaek@dgist.ac.kr )
  • Daehyun Kum(kumdh@dgist.ac.kr)
  • Sang Hyuk Son (son@dgist.ac.kr)