Power Transformer Assessment Based On Evidence Theory

With the construction of domestic EHV and UHV projects, the scale of China’s power grid is expanding. Safe and stable operation of the power system will directly influent the economic interests of the country’s business and users. As the hub of power system equipment, failures on large power transformers will have a negative effect on the whole society. Therefore, the management and the maintenance of power transformers has become an important part of daily maintenance work. It’s very useful to make an effective assessment of power transformers condition for transformer maintenance overhaul and the forecasting of the transformer operation. And it has good influence on the reliability of the power system as well.There are many complex factors affecting the transformer condition. As a multi-attribute decision-making problem, it’s difficult to make an accurate assessment of the power transformer condition. At present, many scholars at home and abroad have studied on the power transformer condition assessment, and achieved some results. This paper has compared the various existing assessment methods at home and abroad, and it has established power transformer condition assessment system according to the relevant regulations and guidelines. On the basis of the system, it has also established a power transformer condition assessment based on the evidence theory.Evidence theory has its shortcomings. Aiming at the rules of evidence theory cannot be used for combination conflict evidence, this paper presents a new approach of evidence combination based on the similarity coefficient between evidence. The new approach still uses the traditional evidence theory combination formula, but calculates the similarity coefficient based on the fuzzy membership of evidence firstly before combination. The basic reliability partition of the evidence is revised according to the similarity coefficient so that the evidence theory can be used for the case of conflicting indexes in the transformer condition assessment, and a more accurate combination result can be obtained. Numerical examples test correctness of the new approach proposed in this paper and point out that it has a certain effect as well.