Positioning of the Moving and Static Contacts of the Switch Machine Based on Double-Layer Mask R-CNN

Abstract

With the continuous development of rail transit, the maintenance of the switch machine becomes more and more important, and the contact depth of the moving contact and static contact in the switch machine is a key part of it. At present, the manual measurement method is the main measure of contact depth, which has the problems of low efficiency and strong subjectivity. The measurement of contact depth based on machine vision includes two steps: moving and static contact positioning and distance conversion. The positioning result will have an important impact on distance measurement. Therefore, a positioning method for moving and static contact based on double-layer Mask R-CNN (DLM) is proposed in this paper: first, the moving contact is roughly positioned by Mask R-CNN to obtain the predicted target area; second, the subgraph of the target area is preprocessed; finally, the precise positioning is used to determine the precise position of the moving and static contact. The accuracy and robustness of the proposed DLM are verified by the internal image of the switch machine.

A Low-Complexity Accurate Ranging Algorithm for a Switch Machine Working Component Based on the Mask RCNN

“…As most recent studies have focused on object detection, few of them present the applications for switch machines. Yin et al [12] attempt to detect the pillar of switch machines with two networks; this is a similar application to our work. Therefore, the algorithm proposed by Yin et al [12] is selected for the comparison.…”

“…Yin et al [12] attempt to detect the pillar of switch machines with two networks; this is a similar application to our work. Therefore, the algorithm proposed by Yin et al [12] is selected for the comparison. The experimental results of the proposed algorithm, Yin et al [12] and Yolo, are shown in Figure 16.…”

“…From Figure 15, it can see that the accurate results are 0.42 mm, 0.46 mm, 0.89 mm, 0.21 mm, 0.17 mm, and 0.3 mm, respectively, which means the proposed algorithm keeps a high level of accuracy in calculating the access depth for the different types of switch machines. It should be noted that the dataset mentioned in Section 3.1 with 2400 images is used for training the networks of Yin et al [12] and Yolo. As most recent studies have focused on object detection, few of them present the applications for switch machines.…”

“…Therefore, the algorithm proposed by Yin et al [12] is selected for the comparison. The experimental results of the proposed algorithm, Yin et al [12] and Yolo, are shown in Figure 16.…”

“…The Mask RCNN [11] adds a new branch network for generating the segmentation mask to the region of interest (ROI). Yin [12] et al improved the positioning accuracy through two Mask RCNN networks. However, the proposed algorithm worked well in the experimental environment.…”