基于支持向量机的火焰状态识别方法

电站锅炉燃烧稳定性的定量判别一直是一个难题。作者首先对火焰图像进行特征提取，提取出火焰亮度、火焰高温亮度、火焰面积、火焰高温面积、火焰高温面积率、质心偏移距离和圆形度等7个特征量。然后分别对这个特征空间和原始图像数据使用支持向量机对其进行识别分类，结果表明：两种数据结果相同．能够正确识别燃烧火焰状态．证明特征量的提取是成功的；支持向量机方法用于燃烧火焰的分类识别是可行的。图4表1参8

Method of Flame Identification Based on Support Vector Machine

The quantitative judgement of combustion stability of utility boiler has been difficult.In this paper, first, the seven features of the flame image are extracted, which were brightness of the flame, brightness of the high temperature flame, area of flame, area of the high temperature flame, rate of area of the high temperature flame, centroidal offset and circularity etc. Then the feature space consisting of seven features and original image data respectively are classified with Support Vector Machine. These two methods drew the same conclusion, and they both can correctly recognise the state of the combustion flame. Further analysis suggested that the feature extract is success and Support Vector Machine is effective for classification of the combustion flame .Figs 4, table 1 and refs 8.