Future failure rate prediction for transformer in power system

• Main Author: Du, Zhijie
• Other Authors: Hu, Guoqiang
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Electrical and electronic engineering::Electric power::Production, transmission and distribution
• Online Access: https://hdl.handle.net/10356/152031
• Institution: Nanyang Technological University

Electric energy has become a necessity in people's production and life. Social development, scientific, technological progress and other aspects are inseparable from electric energy. The power system is an important carrier of power system transmission, distribution and use, which transmits and distributes the power generated by the power plant to each user. The stability of electricity is essential not only for the family but also for the country. The transformer, as the primary equipment for power grade transfer, is responsible for increasing and lowering voltage in the power grid, and is the key node of electric energy transmission. Transformer failure will affect the transmission and distribution of electric energy, causing a large area of blackout with the production and life being forced to shut down, contributing to serious losses for the society and the country. If we can find an effective method to predict transformer faults and take measures in advance for maintenance before transformer faults occur, the probability of transformer faults can be reduced to a large extent. In addition, the transformer fault prediction can also find and solve many potential faults and hidden dangers in time, so that the transformer can operate for a long time. The faults of power transformers are usually shown in the form of light energy, electric energy, heat energy and chemical reaction. In the case of oil-immersed transformers, the gas content in the transformer oil may be used to infer potential faults. As a result, the gases dissolved in transformer oil are primarily acetylene (C2H2), methane (CH4), hydrogen (H2), ethylene (C2H4), and ethane (C2H6). The main sources are the decomposition of solid material and transformer insulating oil. Then, the common faults of transformers are analyzed and three kinds of ways to assess the faults of transformers are given: IEC Three-Ratio method, Japan electric co-research method and Four-Ratio method. After that, the Grey Theory and three generation methods of Grey series operators are introduced: accumulation operator generation, deceleration operator generation and the generation of new information-first weakening buffer operator. The Grey correlation degree is measured and it is discovered that different gases dissolved in the transformer have a coupling correlation. The GM(1,1) model is then created, and a flow chart of GM(1,1) prediction is provided, along with the related Python code. A (1,n) form Grey Model is built according to the (1,1) form Grey Model, and the coupling relationship between various gases in the transformer are taken into account, to better predict the content of dissolved gases in the transformer. Besides, by analyzing the shortcomings of the Grey Theory, four optimization methods are put forward: data preprocessing, background value correction, Grey correlation degree weighting method and equal-dimension new information method to improve the prediction accuracy. Finally, a set of sample data of transformer dissolved gases are selected to validate the prediction of the Grey Model example, as well as the error analysis of the model.