Research On Temperature Characteristics And Hot Spot Temperature Prediction Of Oil-Immersed Transformer

As the society keeping progressive,the demand for electricity in all industries has been growing steadily,so the power system needs to maintain the safe and efficient transmission and distribution state at all times.As one of the pivotal equipment of power system,when it exceeds its service life and is decommissioned or outage maintenance occurs,it will affect the safe operation of the power system.The main reason for this phenomenon is the aging of insulation materials inside the transformer.Transformer in operation due to loss caused by temperature rise will accelerate the aging speed of insulation materials,therefore,monitoring transformer temperature(generally hot spot temperature),for the maintenance of transformer safe and stable operation and thermal structure design guidance has a good practical significance.This paper pays key research on transformer temperature characteristics,and puts forward an algorithm model to predict hot spot temperature,and makes a more in-depth theoretical analysis and experimental verification of related problems.Firstly,various loss mechanisms and temperature rise characteristics of main structural parts(core,winding)of transformer are analyzed,and the influence of different cooling methods on transformer temperature changes is studied.Based on the above research content,the preliminary analysis of hot spots.Secondly,this paper takes a S11-2000 KVA oil-immersed transformer as the research object,establishes its three-dimensional physical model,simulates the transient temperature field of the transformer by FEM software,whose oil circulation is natural,and pays key discussion about the temperature rise characteristics of the transformer from the initial time to the state of thermal balance.The location of transformer hot spot temperature is analyzed qualitatively and quantitatively,and the variation rule of hot spot temperature rise under different conditions is also analyzed.Then,combining the theoretical results of simulation analysis and the demand for hot spot temperature monitoring in realistic engineering,this paper brings foward a hot spot temperature forecast method based on improved BP neural network.Aiming to reducing the complexity of the network,the data samples are pre-processed with the technique of improved principal component analysis method.Taking the shortcomings of traditional BP algorithm into consideration,its initial weight and threshold is affimed with the optimization of the ant colony algorithm.The network behavior and prediction accuracy of BP algorithm are improved by synthesizing the two optimization methods.Finally,the measured data is used to verifying the proposed prediction algorithm,and the prediction results are compared with the guide calculation method,standard BP algorithm,etc.The aspects of network performance,prediction accuracy and evaluation index prove that the proposed algorithm is correct and effective.The research results show that in the transformer model adopted in this paper,the temperature of the low-voltage winding is the highest during the temperature rising,and the hot spot temperature is at the lower position of the top of the B-phase winding during qualitative analysis.The quantitative analysis results show that the longitudinal height of the winding is 0.606 m,accounting for 77.19% of the whole height.When the external temperature and load rate change,the hot spot temperature will also change.The hot spot temperature has a linear relationship with the external temperature,but a nonlinear relationship with the load rate(positive correlation).In this paper the improved BP algorithm has good performance and prediction accuracy than other algorithms mentioned,besides,its mae,mse and mape are 0.0657,0.0067 and 0.44%,respectively.In this paper,the temperature field analysis and hot spot temperature calculation of oil-immersed transformer have been fully studied and verified by experiments,which lays a good foundation for transformer temperature monitoring and thermal structure design in future engineering.