Electric trucks pantograph-catenary interaction condition monitoring method based on semantic segmentation network and linear fitting

Abstract The pantograph-catenary system (PCS) in dual-source electric trucks is crucial for maintaining a stable connection to the power grid, which directly impacts power quality. Ensuring reliable contact between the pantograph and catenary requires accurate detection of the contact point (CPT). However, existing CPT detection methods designed for trains are not well-suited for electric trucks due to differences in structural design. To address this challenge, this paper proposes a novel CPT detection method that integrates semantic segmentation and linear fitting techniques. Firstly, we introduce a lightweight Pantograph and Contact Line Segmentation Network (PCSN), which accurately extracts the regions of the pantograph and contact line. Secondly, a position correction algorithm combined with a least-squares linear fitting technique is employed to detect the CPT of electric trucks. Experimental results demonstrate that the proposed method achieves a detection error within ±5 pixels. In terms of processing speed, it reaches 76.9 FPS on an RTX 3080 GPU, 47.13 FPS on an Intel I9-12900 CPU, and 11.63 FPS on an embedded Jetson TX2 device.