Research On The Prediction Method Of Transformer Top Oil Temperature Based On Bayesian Network

The position of power transformer in power grid system is irreplaceable for many devices.Its safety and reliability is the premise to ensure the safe and stable operation of the power grid system.The top oil temperature is an important factor affecting the service life and load capacity of the oil-immersed power transformer.On the basis of summarizing the research results of transformer top oil temperature,the internal heat transfer process of the transformer is elaborated,and an improved top oil temperature model considering air humidity is put forward.Through the analysis of the influencing factors of transformer top oil temperature,the oil temperature model based on Bayesian network is constructed to predict the top oil temperature of oil-immersed transformer.And the prediction results are applied to the dispatcher's decision in the power grid system,which can obtain certain benefits.The main work of the paper is as follows:Firstly,Through the analysis of complex heating and heat dissipation processes inside the oil-immersed transformer,the internal heat loss is classified,and the flow direction and circulation route of heat are determined,and the heat source of the top oil temperature rise of the transformer is identified.Based on this,deduces the temperature-raising and lowering characteristic equations for windings,cores,and transformer oils.On the basis of the classical calculation model of the top oil temperature and considering the influence of the air humidity in the environment of the transformer,this paper introduces the factor symbolizing the air humidity to build a simple improved model of the top oil temperature.By comparing the calculated values of the model,the calculated values of the semi-physical model and the measured values,it is verified that the oil temperature model which takes air humidity into account has higher accuracy.Secondly,using Bayesian network to predict the top oil temperature of transformer,the construction and reasoning process of Bayesian network model are introduced.Based on the historical monitoring data,the top oil temperature prediction model of Bayesian network is constructed.The comparison results show that the relative error and root mean square error of the prediction model constructed in this paper are 1.0845 and 1.1731,respectively,which are smaller than the other two models,which shows that the prediction effect of this model is more accurate.On the basis of preserving most of the original information,the dimensions of the influence factors of transformer top oil temperature are reduced by principal component analysis method,which reduces the input variables of the model.And finally to establish the Bayesian network model based on principal component analysis.Finally,during the summer peak season,the top oil temperature monitoring value of the transformer is generally high,which leads to heavy dispatching task and many times of load adjustment of power grid.In response to this problem,the top oil temperature prediction value of Bayesian network model is used to assist Dispatchers make decisions to improve power dispatching efficiency.