Impulsive noise in railway automated monitoring: A recursive filtering approach

The inspection of the geometrical proprieties of railways for high speed lines is performed by automated vehicles equipped with laser sensors. The acquired signals are affected by impulse noise, which often causes false alerts on the conditions of the rails. For various technical reasons, it is strongly preferable to remove this noise before the monitored variables are processed and stored. Various dynamic filters have been proposed to cope with this type of noise, including an effective recursive non-linear filtering method based on fuzzy logic. This paper describes a novel variant of the fuzzy recursive filtering algorithm, which improves both spike detection and removal performances. The numerical results are based on a variety of real-world data describing different profiles of high-speed lines in various countries, and highlight the versatility and the effectiveness of the proposed approach