Study Of Dynamic-Overload System Based On Hot Spot Temperature In Oil-Immersed Transformer

The demand for electricity is continually growing with the rapid development of economy, thus there is a problem of power grid that the transmission capacity may not meet the requirement sometimes, our country has drawn out many constructive plans to alleviate this situation. Power transformer serves as one of the most important equipment in the power system, it can run in the state of dynamic capacity-increase because of the hidden overload capacity and this way is the most preferred method to ease the excess demand. When transformers are operating under the condition overload, the inner temperature will rise while the loss of winding will increase, then the isolating remnant life is affected. It is necessary to monitor the hot spot temperature before the dynamic capacity-increase of transformer.Three crucial issues about the implementation of dynamic capacity-increase of transformer were investigated in this dissertation, including the characteristics of temperature rise, the method to predict the hot spot temperature of oil-immersed transformer, investigate the scheme of dynamic overload running and then develop the system named dynamic capacity-increase of transformer based on the research of hot spot temperature.It is significant for the research of hot spot temperature to study the thermal character of transformer. On the basis of realize the structure of oil-immersed transformer, this paper analyzed the process of heat transfer and the inner temperature rise of transformer, then the tendency of winding, iron core and oil was obtained in the course of temperature-rise and cooling stage, the distribution area of hot spot temperature was clear eventually. Three-dimension model of oil-immersed transformer was established by the finite element software, the distribution of temperature can be obtained through the accurate calculation and analysis on the coupling field of magnet-fluid-thermal, the conclusion of electromagnetic analysis was generated into the thermal field model as the pre-additive load in the thermal analysis, verifies the conclusion of the thermal characters studied before and a deeper comprehension of hot-spot temperature is obtained simultaneously, which provided theoretical basis about the installation of optical fiber sensor in measuring temperature.On the basis of knowing the inner distribution of transformer and the location of hot spot temperature, the neural network was used to predict the hot spot temperature of transformer. Several issues were studied including selection of the training parameters, structure design of neural network, processing of training sample and algorithm improvement of neural network while the experimental data was selected as learning sample. In the process of establishing network, the output dimension reduction method was adopted and the Levenberg-Marquardt algorithm was used to improve the network, the improved three-layer network was established eventually. After the training of neural network which database derived from three transformers, these data was divided into two parts with one named training set and the other was called validation set, neural network was validated after the training process, this chapter compared the values of improved neural network, IEEE STD C57 guide recommend method and actual measuring, verifies that the improved neural network has the best performance in predicting the hot-spot temperature of transformer.With the restriction of national standard about the maximum allowable hot spot temperature of transformer, considering the life loss of transformer at the same time, this article researched the safe operation of transformer under the short-term emergency load and long-term emergency load, then synthesize the study of hot spot temperature, this paper developed the dynamic capacity system of transformer eventually, after the hot spot temperature was predicted by the use of real time operating data, this system can calculate the maximum allowable operating factor on the premise of load type was distinguished, it provides load-distribution evidence for the scheduling department. Three aspects are introduced including the implementation process, technical characteristics and functions, the benefits of overload is also analyzed and the decisions driven by the system was ultimately proved to be valuable.