Railway's Turnout Fault Diagnosis Based on Power Curve Similarity

Aiming at the problem of complex process, time-consuming and low accuracy of traditional turnout fault diagnosis methods, a turnout fault diagnosis method based on similarity degree of turnout power curve is proposed. Firstly, according to the matching degree with standard curve, the fault curve is divided into easy-to-find fault and difficult-to-find fault. Referring to microscopic idea, rough similarity is used to deal with easy-to-find fault and fine similarity is used to deal with difficult-to-find fault. At the same time, the rough similarity index based on Hausdorff distance and the fine similarity index based on dynamic time bending are designed respectively. The power curve of S700K turnout collected by Lanzhou West Railway Station is simulated to verify the feasibility of this method. The accuracy and time-consuming of this method are compared with those of traditional BP neural network and SVM methods, which proves the superiority of the curve similarity method.