Abstract: Fires of oil immersed transformers may cause serious consequences. A dynamic fire risk assessment method was suggested to provide guidance for the prevention and risk management of transformer fires. In this paper, the causes of oil immersed transformer fire are analyzed based on accident cases and industry statistical data. A transformer fire risk index system covering six dimensions including the inherent risks of transformer components, dynamic operation risks, failure risks of ancillary facilities, external environmental factors, reliability of firefighting equipment, and fire safety management level has constructed. A single index classification evaluation method covering four severity levels has proposed. The chain evolution model of transformers fire has developed using complex network method, the index weights are calculated based on the measurement method of node importance. A calculation method for the overall fire risk value and extremely fire risk value of transformers is proposed based on the TOPSIS method, and the application process of the evaluation model is demonstrated through practical examples. The results show that oil temperature rise, short circuit and arc discharge are the most important fire risk points of transformers. The proposed transformer fire risk assessment method can further refine and present the risk conditions of each dimension on the basis of providing the overall risk level. The research results can provide guidance for the dynamic monitoring of transformer fire risks and the implementation decision-making of risk reduction measures.

Key words: transformer fire, chain evolution, complex network, risk assessment, TOPSIS method