空气预热器热点检测系统的研究

针对现场环境,用低照度CCD获取空气预热器金属部件在不同温度下图像信息,通过小波包变换,以最佳子空间下每层熵值总和作为特征向量元素,按分解顺序排列每层熵值总和,获得图像的特征。采用具有距离特征的类别可分性判据,对空气预热器不同的工作状态(正常、发生火情)进行识别,试验表明,该方法准确有效。图3表1参7

An Air Preheater's Hot Spot Detection System

A low-luminosity CCD camera is being used to catch image information of air preheater elements distinguished by different temperatures, which, by wavelet packet transformation and summation of each layer's entropy values in optimal subspaces, are defined as feature vector elements. The characteristics of the images are observed by the summed up entropy values of each layer, arranged according to the sequence of decomposition. With the help of the classification separability criterion with distance characteristics, different operational states (normal, on fire) can then be distinguished. Tests show that this method is veracious and effective. Figs 3, table 1 and refs 7.