Research On Temperature Rising Prediction Of Distribution Transformer By Artificial Neural Networks

Transformers play a pivotal role in the process of power transmission and distribution.The service life of the transformers generally depends on the life of insulating materials.The high temperature is the main reason that results the aging of the insulation material.We can predict the service life of a transformer by monitoring the hot spot temperature of transformer winding.The research on the hot spot temperature of transformer winding can also guide the design of transformer theoretically.Therefore,studying the thermal characteristics of transformers has high theoretical value and practical value.In this paper,the temperature rise of transformer winding hot spot is studied,and the change rule of transformer winding hot spot temperature rise is explored by using the method of neural network.The main tasks are as follows:(1)This paper investigates the research status of transformer temperature rise at home and abroad,and analyses the principle of heat source inside the transformer.According to the experimental data measured in practice,the influencing factors of transformer temperature rise are determined.Under the condition of single factor,the influence of P?Q ?Vthd?Ithd?Tiand Tiron-1on transformer temperature rise T is explored.(2)This paper introduces the basic theory of neural network,including the model of biological neural network,the mathematical model of neural network,the topological structure of neural network,the principle of BP network,and the logic process of specific training method Traincgp.(3)Active power(P),reactive power(Q),total voltage harmonic distortion(Vthd),total current harmonic distortion(Ithd),room temperature(Ti)and the temperature of power distribution transformer(Tiron-1)in the first half hour are used as input variables of the neural network.The output terminal is the predicted temperature of the power transformer.According to the change of load,the predicted temperature rise of the transformer is divided into three parts,the average workingday,Saturday and Sunday,and the temperature rise of transformer is predicted by stages,the variation of error is analyzed according to the actual application scenario.(4)In order to explore the influence of harmonic factors on transformer temperature rise prediction,on the basis of six factors,the first half-hour voltage harmonic factor,the first half-hour current harmonic factor and the first half-hour room temperature are added to predict the transformer temperature rise.At the same time,the prediction results of six factors and nine factors were compared.The experimental results show that the average prediction error based on six factors is slightly smaller than that based on nine factors except for the low load area and the load stationary area of the working day.In the other cases,the average prediction error based on nine factors is less than that based on six factors.Therefore,under the condition of the selected influence factors in this paper,the prediction of temperature rise based on nine influence factors is better than that based on six influence factors.