Research On Transformer Fault Diagnosis Based On Digital Twin Technology

In the power system,oil-immersed power transformers carry the functions of voltage conversion and electric energy transmission,which is the key equipment to ensure the safe operation of the power grid and the safety of the people using electricity.With the rapid development of China's economy in recent years,the demand for electricity is increasing,and the requirements for guaranteeing industrial electricity,residential electricity and commercial electricity are getting higher and higher.Therefore,it is an important task to ensure the safe and reliable operation of power transformers.Currently,among transformer fault diagnosis methods,the most commonly used is Dissolved gas analysis(DGA)method.However,such methods are usually based on human experience judgment,and have problems such as absolute fault classification,inaccurate judgment results,and insufficient diagnosis speed.When transformer fault monitoring is carried out,traditional 3D model is generally adopted,which cannot realize the information interconnection between physical space and virtual space and carry out real-time online monitoring.Digital twin technology as information Physical Systems(CPS,Cyber-Physical Systems)of key technologies,with the development of big data,intelligent algorithm,has realized the landing in the field of more and more application,but at present is still in the preliminary exploration for transformer fault diagnosis,stage,if able to build the perfect digital transformer twin model and combining with fault diagnosis intelligent algorithm,can make it possible to transformer on-line monitoring.In view of the above problems,this paper focuses on the research method of transformer fault diagnosis based on digital twin technology.The main research contents are as follows:(1)Research on transformer fault diagnosis architecture and digital twin.The transformer structure and operation characteristics are analyzed,the transformer fault diagnosis system architecture is built based on digital twin technology,and the concept,composition and main characteristics of the five-dimensional model are described.On this basis,the digital twin model is established to study the operation mechanism of transformer fault diagnosis,which provides general guidance and theoretical basis for the realization of real-time online monitoring of transformer running state in the following paper.(2)Study on transformer fault diagnosis based on optimized probabilistic neural network.The characteristics of gas data in transformer oil are analyzed and the differential evolution algorithm is adopted to optimize the probabilistic neural network for transformer fault diagnosis.Firstly,the two intelligent algorithms are briefly described,and the optimization process and detailed steps are designed.The superiority of the algorithm is verified through simulation experiments.Finally,the optimization algorithm is compared with other neural networks.(3)Application of transformer fault diagnosis based on digital twin technology.A transformer fault diagnosis system is established based on the transformer fault diagnosis architecture and digital twin model.The gas number in the transformer oil is uploaded to the twin database through the data acquisition and transmission device.The fault diagnosis results are diagnosed through the fault diagnosis model in the digital twin,and the results are displayed on the application interface of the system finally.