基于卷积神经网络的智能变电站设备故障检测方法研究

打开文本图片集

中图分类号：TM732 文献标志码：A 文章编号:1003-5168（2025)20-0024-04

DOI:10.19968/j.cnki.hnkj.1003-5168.2025.20.005

Research on Intelligent Substation Equipment Fault Detection Method BasedonConvolutionalNeuralNetwork

GAOYuan JIANGJiafu (State Grid Shanghai Fengxian Electric Power Supply Company,Shanghai 2O1499,China)

Abstract:[Purposes] Intelligent substations have adopted a large number of new types of equipment and components. Current fault diagnosis methods that still rely on shallow image features lead to a decrease infault detection quality.To address this,a fault detection method for intelligent substation equipment based on a convolutional neural network is proposed.[Methods] Firstly,an improved non-local mean filtering algorithm is used to processthe infrared images of inteligent substation equipment,aiming to recover the effective detail information contained in the original infrared images to the greatest extent.Secondly,based on the grayscale image corresponding to the filtered infrared image,the image contrast is enhanced through top-hat transformation and superposition operations.Then,the watershed algorithm is employed to segment the suspicious fault regions in the image.Finally,a fault detection model is constructed with the convolutional neural network as the core.The segmented images of suspicious regions are input into this model to extract shalow visual features and deep semantic features.The classification and decision-making performance of the fully connected layer is leveraged to obtain the equipment fault detectionresults.[Findings] The proposed method maintainsanF1-score above O.9 for inteligent substation equipment fault detection results,achieving accurate diagnosis of equipment faults.[Conclusions] This method not only improves the accuracy of fault detection but also reduces reliance on manual experience,thereby enhancing the automation and intelligence level of diagnosis.

Keywords: intelligent substation; convolutional neural network; filtering processing; image segmentation; feature extraction; fault detection

0 引言

智能变电站作为电力系统的关键组成部分，不仅承担着电力传输与分配的重要任务[]，而且集成了许多先进的信息技术，实现了对设备运行状态的不断监控。（剩余6197字）