作　　者： (梁玲）;

机构地区： 广西现代职业技术学院,广西河池547000

出　　处： 《中国有色冶金》 2017年第4期24-27,共4页

摘　　要： 长期以来国内外对烟化炉还原终点的判断,主要是现场工人观察,判断的准确性受主观因素影响较大。本文研究了在MATLAB平台上,借助于工具箱中的图像处理函数,对三次风口火焰图像预处理,提取火焰特征量,进行神经网络设计与训练,最后进行仿真测试。实践中运用该方法,还原终点识别效果较好。 For a long time, the judgement of reduction ending point of fuming furnace both at home and abroad has been mainly depending on manual observation on-site, so the accuracy is influenced by subjective factors. Based on the MATLAB platform, by means of image processing function in the toolbox, this paper studies on the preprocess- ing of three times flame images at air outlet, extracts the flame features, carries out neural network design and training, and finally develops simulation test. It is concluded that this approach has better effect on the identification of reduction ending point in practice.

关 键 词： 烟化炉 还原终点 图像处理 图像特征 神经网络 网络