A fuzzy logic based filter for spike-noise detection in railways monitoring systems

This paper describes a fuzzy approach for the real-time detection and removal of impulsive spike-like noise from measures for track geometry in railway infrastructures. Spike noise is caused by random events, including anomalous reflections of external lights, detection or transmission errors within the monitoring system, singular parts of tracks (switches, level crossings) hiding the rail profile. Its effects are extremely undesirable, since it can generate erroneous alerts about inexistent railway defects. The real-time detection of such anomalies is a difficult task, due to the high variability of the detected signals, and to the dynamic conditions of measurement. By incorporating the typical rules used for off-line analysis of the measured data in the fuzzy system, we develop a filtering algorithm that is able to perform a spike removal in real-time. Some experimental results are provided to establish the effectiveness of the proposed approach, which has been recently implemented on some measuring vehicles for high speed lanes.