基于托辊裸露特征的输煤皮带跑偏检测方法OA

To achieve real-time and accurate detection of belt deviation in belt conveyors under com-plex industrial environments,this paper proposes a method that integrates an improved lightweight seman-tic segmentation model based on YOLO11 with an engineering-friendly deviation criterion utilizing idler exposure areas.The improved YOLO11 model incorporates a C3k2-Edge module in its backbone,which enhances edge perception capabilities in complex backgrounds through multi-scale edge feature extraction and a DSM attention mechanism.The neck is designed with a HS-PAN network that integrates top-down and bottom-up paths to improve feature fusion and localization efficiency.Results from experiments on the self-built dataset and actual deviation detection scenarios demonstrate that the proposed method can meet the requirements of high-precision,lightweight,and real-time belt deviation detection in com-plex industrial settings,exhibiting good engineering applicability.