Pantograph is an important part of the metro traction power system. In the long-term operation of the metro, the pantograph carbon slider continues experiencing wear. If the pantograph wear detection is not timely or accurate enough, it is easy to cause failure and even metro safety accidents. Therefore, it is important for the safe operation of the metro to realize high-precision online detection of pantograph wear. At present, the metro pantograph wear detection is mostly manual detection and image detection. The manual detection method requires metro shutdown, and the detection efficiency is low. The image detection method can realize the non-contact online detection of pantograph wear, which improves the detection efficiency and accuracy. However, the image detection method is affected by factors such as illumination, vehicle speed, and interference. There is a situation that the quality of the original picture is poor and the pantograph profile cannot be accurately extracted, which greatly lowers the accuracy of the pantograph wear detection. Aiming at the pantograph wear detection in metro operation, an online detection method of metro pantograph wear based on line-laser measurement is proposed to realize high precision, high efficiency and online detection of metro pantograph wear.
In this study, the line-laser sensor is used to realize the data acquisition of the pantograph profile of running metro. For the collected pantograph profile data, firstly, the effective pantograph profile data are selected by threshold setting. Secondly, a feature point search algorithm based on point cloud window moving calculation is proposed to realize the feature point search of pantograph profile data and separate the carbon slider profile data. Then, the Savitzky-Golay filtering algorithm is used to denoise the carbon slider profile data. Finally, the combination of coarse registration based on principal component analysis (PCA) and accurate registration of iterative closest point (ICP) algorithm with interval constraints is used to realize the registration of carbon slider profile and standard profile, and the evaluation of pantograph wear is completed. The accuracy and effectiveness of the feature point search algorithm based on point cloud window moving calculation and the pantograph wear detection method are verified by experiments.
In order to realize the detection of pantograph wear in running metro, an online detection system of pantograph wear in metro based on line-laser measurement is designed, and the data of pantograph profile are collected by line-laser sensor (Fig.3). The effective profile data of pantograph are screened out by setting thresholds for the height and number of pantograph profile data, and the effective profile data of pantograph are screened out (Fig.4). The feature point search algorithm is used to solve the minimum variance index and the maximum point-line distance of the profile data (Figs.7 and 8). The accuracy of the feature point search algorithm is verified by four sets of collected pantograph profile data (Fig.9 and Table 1). The Savitzky-Golay filtering algorithm is used to denoise the separated carbon slider profile data (Fig.10). The initial position of the separated carbon slider profile and the standard carbon slider profile is adjusted by the rough registration based on PCA (Fig.11). The profile intervals involved in accurate registration are divided, and the final registration result is obtained by ICP to complete the pantograph wear evaluation (Figs.12 and 13). The accuracy of the detection method is verified by the pantograph wear detection experimental platform, and the wear values of five positions of the pantograph are detected and compared with the standard values (Table 2).
In this paper, an online detection method of metro pantograph wear based on line-laser measurement is proposed to achieve the accurate and efficient detection of running metro pantograph. The detection method realizes the collection of running metro pantograph through line-laser sensor. Firstly, the collected pantograph profile data are preprocessed by threshold setting, and the effective pantograph profile data are selected. After preprocessing, a feature point search algorithm based on point cloud window moving calculation is proposed. The feature point search algorithm finds the critical feature point of the horn profile and the carbon slider profile by traversing the profile data, and separates the carbon slider profile data. The Savitzky-Golay filtering algorithm based on least squares is used to denoise the separated carbon slider profile data to eliminate the interference of noise points on profile registration. Finally, the combination of coarse registration based on PCA and accurate registration of ICP with interval constraints is used to realize the registration of carbon slider profile and standard profile. According to the registration result of carbon slider profile, the accurate evaluation of pantograph wear is realized. The effectiveness of the pantograph wear detection method is verified through the detection experimental platform for pantograph wear, and the detection method meets the wear detection requirement of ±0.5 mm.
