【摘要】 针对变电站指针式仪表尺寸较小难以识别的问题，该文提出改进YOLO11n的变电站指针式仪表检测算法。引入可重参数化重聚焦卷积（RefConv）强化特征提取能力；分组混洗卷积（GSConv）降低计算复杂度；采用空洞空间金字塔池化模块（ASPP）扩大网络感受野的同时提高多尺度感知能力；利用可伸缩的IoU损失函数（SIoU）对模型训练进行监督，实现对变电站场景中的指针式仪表的准确检测。实验表明，所提算法检测精度达96.9%，相较于原始YOLO11n算法，精度提升3.5%，参数量下降4.5%，有效平衡了精度与实时性需求。该方案为变电站智能巡检系统提供了高精度、低延时的检测方案，对推动电网设备智能化监测技术发展具有重要意义和价值。更多还原

【Abstract】 To solve the problem of detecting small-sized pointer-type instruments in substations,this study proposes an enhanced YOLO11n-based detection algorithm. Ref Conv is introduced to boost feature extraction. GSConv is used to cut computational complexity. The ASPP module expands the network’s receptive field and strengthens multi-scale perception. SIoU loss supervises model training. Experiments show the algorithm achieves 96.9% detection accuracy,a3.5% improvement over YOLO11n,with a 4.5% reduction in parameters. It balances accuracy and real-time performance,offering a high-precision,low-latency solution for substation inspection systems and advancing grid equipment monitoring tech.更多还原