Burner Flame Recogniton Based on Backpropagation Neural Network
基于BP网络的锅炉炉膛火焰燃烧状态自动识别

In this paper, a system based on ANN for flame recognition and extinguishing alarm for a boiler is presented. We use CCD camera, fibre and image card to grab the flame image, then identify its burning state using ANN after preprocessing. According to the result of identification the system decides to alarm for extinguishing or not. The ANN is trained with the backpropagation learning algorithm. The experiment result shows that the system has high correctness of flame recognition. It is effective to alarm for flame extinguishing.