Research on Contour Recognition Methods for the Moving Contact of Miniature Circuit Breakers Under Arc Interference

Abstract This paper addresses the challenge of detecting moving contacts in miniature circuit breakers (MCBs) under strong arc interference during the breaking operation by proposing an object detection framework that integrates arc segmentation with an improved YOLOv11 model. First, YOLOv11 is employed to locate the region of the fixed and moving contacts, and an arc constraint region is constructed based on their spatial relationship. Then, an adaptive fuzzy C-means (AFCM) clustering combined with a locally constrained implicit level set (LICLS) method is applied to achieve coarse-to-fine segmentation and image inpainting of the arc region, thereby mitigating the influence of intense arc light on subsequent moving-contact detection. Finally, a YOLOv11-SDAU model is developed by incorporating a Spatial Enhancement Attention Module (SEAM), a Deformable Large Kernel Attention (D-LKA) mechanism, and the UIoU loss function to enhance detection capability under complex occlusion scenarios. Experimental results demonstrate that the proposed method achieves high detection accuracy and robustness under strong arc interference, with a model [email protected] of 95.18% and a 3.37% improvement in [email protected]:0.95 over the baseline model. Furthermore, by integrating high-speed imaging with electrical waveform data, time parameters such as the drop-out time are measured and analyzed. The findings of this study can serve as a reference for condition monitoring and performance evaluation of miniature circuit breakers.