Intelligent Inspection Technology for Substations Based on Multimodal Sensing
To tackle the inefficiencies and reliance on manual labor in the identification and qualitative fault detection of substation equipment,this paper proposes an enhanced YOLOv5 model for the automated recognition and fault detection of substation devices.It firstly introduces ShuffleNet v2 into the backbone network to reduce the computational complexity and parameters so as to facilitate lightweight processing.Subsequently,it introduces the efficient intersection over union(EIOU)loss function to refine the regression accuracy and expedite the convergence of predicted bounding boxes.Finally,a convolutional block attention module(CBAM)is embedded within the network to improve the model's detection accuracy.Comparative experiments conducted on a custom dataset show that the improved YOLOv5 model outperforms six other models in terms of parameter volume,computational amount,and average accuracy.The ablation studies further verify the contributions of ShuffleNet v2 and the CBAM module in enhancing detection precision and real-time performance.These enhancements lead to a reduction of 5.26 Mibit in parameters and 10.3 Gibit decrease in computational amount compared to the original YOLOv5 model,along with a 4%increase in average accuracy,indicating the model's potential for application in the intelligent inspection of substation equipment.
substation equipmentYOLOv5ShuffleNetconvolutional block attention module(CBAM)efficient intersection over union(EIOU)
万方数据
维普
