关键词: 无人值守变电站, 数据融合, BP神经网络, 模糊推理

Abstract: In view of the high false alarm rate in the traditional fire alarm system of substation, it is impossible to take fire alarm and fire protection measures with different degree of strict degree according to the importance of different areas in the substation, and proposes a fire detection algorithm of unattended substation based on data fusion technology. In the feature layer of data fusion technology, BP neural network is used to fuse the temperature, smoke and CO in the detection area, and the probability of open fire and smoldering fire is predicted; In the decision-making layer, the fire probability output by the feature layer is combined with the three additional information, including fire duration, fire risk and damage degree, and finally the fire alarm level is output. Finally, the simulation results show that the algorithm can identify the scene of the burning fire and smoldering fire quickly and accurately, and can give reasonable alarm decision-making according to the importance difference of different detection areas. It has certain flexibility and advanced nature.

Key words: unattended substation, data fusion, BP neural network, fuzzy inference