基于图像技术和PCA方法的火焰监测研究

Research on Flame Monitoring Based on Image Processing Technology and PCA method

从火焰图像提取了能够反映火焰燃烧状态的7个特征量:火焰亮度、高温区域亮度、火焰面积、高温区域面积、高温面积率、火焰质心偏移距离和火焰区域的圆形度。然后,基于主元分析技术,提出一种对燃烧火焰的稳定性进行了监视和诊断的方法,采用HotellingT2和SPE两个统计量对每一时刻的图像数据向量进行监测,检验是否超过各自的控制限,只要这2个统计量之一越限,则可判定燃烧出现异常,并且将结果以SPE图、HotelliingT2图和主元图的形式直观地表示出来。

The seven characteristic values of flame luminance, high temperature flame luminance, flame area, high temperature flame area, high temperature ratio, centroid offset distance and circularity etc. are picked up by analysis ono flame image. Then a method to monitor and to diagnose the stability of burniing flame is put forward base upon PCA (principal component analysis) method, it uses the two statistic of Hotelling T2 and SPE monitor the every-time image data vectors to check them whether they exceed their own controllable limit. As long as one of them exceeds the limit, abnormity combustion can be concluded. The result is showed with SPE fig, Hotelliing T2 fig and PCA fig visually. Figs 5, table 1 and refs 8.