Research On Fault Analysis Of Power Epuipment Based On Infrared Image

With the rapid development of China’s comprehensive national strength and the increasing dependence on power,the importance and necessity of the power industry is self-evident.Whether the power equipment can operate safely,reliably and stably for a long time is the key to the whole power system.Therefore,in order to ensure the safety and reliability of the operation of power equipment,it has become a research hotspot to monitor the power equipment effectively,observe and record its temperature changes,analyse the power equipment for fault degree,fault location,fault prediction in real time and automatically.In this paper,relied on infrared thermal imaging technology,two methods based on deep learning are designed to detect and analyze the fault of power equipment.A fault analysis method for infrared image of power equipment based on convolutional neural network is proposed.First of all,image preprocessing is performed.According to the characteristics of infrared image noise in power equipment,a method of combining mean filtering and median filtering is adopted.The comparison proves that the method has a good denoising effect,which lays the foundation for subsequent image segmentation and fault localization;secondly,the convolutional neural network model for fault classification of power equipment infrared image is designed.The experiment shows that the network has extremely high recognition ability and is suitable for actual production;Finally,fault analysis and health management are carried out for fault power equipment.Fault analysis includes fault location and fault level judgment.Fault area location depends on image segmentation technology.This paper proposes an algorithm based on pixel clustering,which combines SLIC and global threshold.Experiments show that the algorithm not only has good segmentation effect,but also has label function,so as to realize the real localization of the region and facilitate the subsequent temperature extraction operation,combined with a certain applicability of the relative temperature difference criterion to achieve fault level judgment.In addition,surrounding the idea of health management,the spatiotemporal characteristics of infrared image sequence are introduced and the fitting method is used to predict the equipment state trend curve.The GUI interface is designed to complete the centralized display,classified storage,and unified management of faulty power equipment,which will lay the groundwork for future research,improve the diagnostic efficiency and accuracy of the infrared image of power equipment simultaneously.A fault analysis method for infrared image of power equipment based on SSD_Mobilenet network is proposed.This method is the attempt to detect fault target in infrared image of power equipment under the rapid development of computer vision,deep learning and target detection technology.Only the feasibility experiment was conducted based on the existing limited data,and the entire analysis process is given in this paper:Based on the basis of classified storage in Method 1,after obtaining enough sample data sets and using the SSD_Mobilenet target detection network for processing,the fault location and defect level of power equipment can be obtained directly.Finally,the health management content including trend prediction and centralized display be performed.So as to achieve a higher degree of automatic detection,further simplify detection steps and improve the detection efficiency in production.