Fault Diagnosis Of Oil Immersed Power Transformer Based On Long Short Memory Network

With the rapid development of the economy,the scale of construction of China’s power grid has become larger and larger.Substation equipment is one of the key equipment for the grid to ensure the user’s electricity consumption.The power transformers of different voltage levels serve as the power supply tasks for different users in the station area.Oil-immersed power transformers are under high-load operating conditions for a long time,and it will inevitably fail over time.Due to the failure of the main transformer of the substation,the power transmission line side users under the substation have power outages,which can lead to large-scale power outages in severe cases.At present,the overhaul of oil-immersed power transformers is mainly planned maintenance,and it has not really achieved state maintenance.Therefore,the safe and stable operation of the oil-immersed power transformer is also one of the keys to the long-term stable operation of the entire substation power system.In this paper,the fault diagnosis of oil-immersed power transformer based on long-term and short-term memory network is discussed.According to the current situation and existing problems of intelligent monitoring and fault diagnosis of oil-immersed power transformers,the following research is carried out:(1)Through a large number of reading domestic and foreign literatures and summarizing field experience,research and summarize the intelligent online monitoring and fault diagnosis methods corresponding to the main fault categories of oil-immersed power transformers,and apply the long-short-term memory network algorithms in the electrical field and other fields.The situation has been thoroughly studied and summarized.(2)The basic content of the main fault categories of oil-immersed power transformers and the latest detection techniques are studied.The main fault categories of oil-immersed power transformers and the latest monitoring and diagnostic techniques corresponding to different types of faults are studied to obtain the actual characteristic parameters.(3)The fault diagnosis model of oil-immersed power transformer based on long and short time memory network is studied in detail,including the overall design of fault diagnosis model,the acquisition of various characteristic data of oil-immersed power transformer and oil immersion based on long and short time memory network.Power transformer fault diagnosis algorithm.The fault diagnosis algorithm model of oil-immersed power transformer based on long and short time memory network is established,including model characteristics,construction of hierarchical long-term memory network and evaluation criteria of diagnostic algorithm.(4)Research on the realization and training of the diagnostic model based on Lab View and Python for long and short time memory networks,and study the overall design of the software,including user login,data acquisition,data analysis,and data query.(5)Through the case analysis,the simulation training of oil-immersed power transformer fault diagnosis algorithm based on long-term and short-term memory network was carried out.A certain volume of data samples was established,and the data diagnosis and algorithm accuracy rate diagnosis were carried out to verify the oil.The effectiveness of the immersed power transformer fault diagnosis decision-making model.