Fault Diagnosis Of Oil-immersed Transformer Based On Convolutional Neural Network

After entering the new century,my country’s economic volume has increased rapidly.At the same time,people have become more and more dependent on electricity,and the safety of electricity has also been concerned by relevant government departments.Because the transformer is the most critical part of the power system,the normal operation of the power system is closely related to its reliable work,so it is very necessary to diagnose the fault of the transformer.In the past,the equipment maintenance mode of the power system was "regular maintenance",that is,to perform maintenance at a fixed time,but now it has become a more scientific and reasonable "condition maintenance".It has been determined that the Dissolved Gas Analysis(DGA)technology in transformer oil is an effective "condition maintenance" mode and an important method for transformer fault diagnosis.Therefore,the data used in this article is derived from DGA.Aiming at the disadvantages of existing fault detection methods such as low accuracy,poor convergence,and easy misjudgment,this paper designs a fault diagnosis model for oil-immersed transformers based on convolutional neural networks based on the rapid development of artificial intelligence technology.Starting from the common faults of oil-immersed transformers,this article first introduces the research background,significance and current research status,analyzes the fault types of transformers based on the referenced literature,and clarifies the causes of gas in the transformer oil and the faults in the oil and Characteristic gas connection.Secondly,the concept of neural network is introduced.Starting from the BP neural network,the BP model is constructed and trained and simulated,and the recognition rate is 90.5%.In order to further improve the fault recognition rate and accuracy,the concept of deep learning is introduced,and the basic principles of convolutional neural networks,as well as the structure and working mechanism of the network are explained.Based on this,a model of convolutional neural network is designed and DGA data is used.Do network training and result analysis,the recognition rate can reach 95.2%.Then compared the two models and found that the convolutional neural network is better than the BP neural network.Finally,two models were used to make a comprehensive diagnosis on the faults of a single transformer and multiple transformers,and the ideal results were obtained.Through the research of this article,the previous transformer detection method is improved,and the application of deep learning convolutional neural network in fault diagnosis greatly improves the detection accuracy and accuracy of transformer fault diagnosis using DGA as data,which is helpful to ensure the safe operation of transformers and it has a positive impact on cost reduction.Figure [19] Table [23] Reference [59]...