Parametric Synthesis of a Track Condition Classifiier
Е. М. Tarasov, И. К. Андрончев, А. А. Булатов, Valery B. Teplyakov · Russian Electrical Engineering · 2020
Abstract An approach to the synthesis of track condition taught classifiers of the main railways is considered. In contrast to the conventional synthesis of rail chains with two classes of states, it is proposed to divide the track state into three classes, distinguishing a control mode class, which allows reliable identification of a rail break. An algorithm for classifier parameters optimization is developed, the research results are presented. The proposed technique can significantly increase the depth of recognition and classification of the track state space.