Abstract: Electrical fires in the early stage of ignition are often characterized by concealment, slow development cycles and high hazard. It is particularly important to find suitable early warning methods to effectively and accurately prevent electrical fires. This paper introduces the causes of electrical fires and the differences between electrical fire early warning and traditional fire early warning. It elaborates on the research status of three electrical fire early warning methods: very early fire detecting system, multi-sensor information fusion and artificial neural networks. The study focuses on the electrical equipment in substations for fire early warning analysis, providing feasible research ideas and methods for the construction of integrated fire warning system for substations.

Key words: electrical fire, early warning method, multi-sensor, artificial neural networks, substation