Study On The Anomaly Detection System Of The Fingerboard For Straddle-type Monorail Traffic Based On 3D Camera

As a standard of urban rail transit,straddle-type monorail has the advantages of lightness,fast,and environmental protection,it has practical significance for solving the increasingly serious traffic congestion problem.Due to the long-term load bearing and the impact of the vibration generated by the train operation,the track beam is unavoidably abnormal.At present,it mainly relies on manual detection,which has low efficiency and poor accuracy,and cannot ensure the safety of rail transit operation,which brings hidden dangers to the safe travel of citizens.Therefore,it is of great practical significance to develop a cross-seat monorail track health detection system.Among them,the abnormal detection of the finger plate is an important part of the monorail track health detection.This paper analyzes and studies the collection and analysis of point cloud data,and proposes a 3D camera-based finger plate anomaly detection method,which can achieve accurate,fast,safe and efficient collection of 3D beam surface data,and can automatically analyze and find The abnormal situation of the finger plate has important practical value for the safe operation of the straddle-type monorail.This article mainly does the following work:(1)The detection system scheme based on three-dimensional camera is proposed.According to the fingerboard object and the field working environment,the lens,line structure light source,camera,photoelectric encoder and other devices are analyzed and selected.Build a three-dimensional data acquisition system of straddle-type monorail track.According to the actual situation,calculate and determine the geometric relationship and relative position of the camera system through experiments to ensure the accuracy and quality of the collected images.(2)In the development environment of Qt5.9,C ++ language is used to write the data acquisition program of straddle-type monorail track,and the camera is calibrated using the traditional three-dimensional calibration method to achieve accurate and high-speed point cloud data collection,and finally use SQLite database storage.(3)Pre-process the point cloud data,according to the characteristics of the point cloud data of the finger plate,a point cloud segmentation algorithm based on geometric feature statistics is used to achieve point cloud positioning,so as to segment the beam face and the finger plate;for the common finger plate Two abnormal conditions:staggered and loose fasteners.The stagger detection method based on plane fitting and the anomaly detection algorithm for fasteners based on point cloud registration are respectively proposed to realize the abnormal detection of finger plates.The experimental results show that the 3D camera-based cross-seat monorail traffic finger plate anomaly detection system has an acquisition accuracy of ± 0.15 mm and is robust.The data anomaly detection system can effectively detect the 2mm stagger of the finger plate.The detection accuracy of the finger plate bolt looseness is 0.3mm.The realization of anomaly detection will provide technical support for the safe operation of straddle-type monorail traffic.