摘要
当可燃气体浓度含量达到一定程度时,遇到火源就会引起爆炸造成经济损失。通过对可燃气体爆炸风险的特征变化能预测在浓度达到极值前,进行危险判断。提出基于对痕量多组分可燃气体浓度变化图谱特征提取改进算法,对可燃气体浓度图谱特征进行处理,获得归一化后的图谱灰度图像信息;利用局部而知模式提取Gabor图谱纹理特征;最后通过迭代计算采集后的特征。利用卷积运算对不同气体的特征进行分类,利用分类后的结果实现痕量多组分可燃气体浓度的图谱识别。实验证明运用气体图谱特征改进算法对可燃气体浓度特征提取后,通过卷积运算对痕量多组分气体浓度特征识别,能预测出可燃气爆炸的风险趋势变化。
When the concentration of combustible gas reaches a certain level, it will cause economic losses when the fire source is met. According to the risk of explosion of combustible gas, the characteristic change of the gas explosion risk can be predicted before the concentration reaches the maximum value.Put forward the improved extraction of trace components of combustible gas concentration profiles based on the characteristics, processing of combustible gas concentration map features, information map of the gray- scale images is normalized; using local knowledge pattern extraction Gabor fingerprint texture features; finally through iterative calculation after collection. Using convolution operation to classify the characteristics of different gases, using the results of the classification to realize the identification of the concentration of trace fractions of combustible gas. The experiment proves that the improved algorithm for extracting features of combustible gas concentration after using gas chromatograms, through the convolution of trace multi component gas concentration feature recognition, can predict the risk of gas explosion change trend.
作者
周作梅
Zhou Zuomei(Information Engineering College KaiLi University,KaiLi Guizhou 556011,China)
出处
《科技通报》
北大核心
2016年第12期151-154,共4页
Bulletin of Science and Technology
基金
贵州省教育厅自然科学基金青年项目资助(黔教合KY字[2015]421)
关键词
痕量多组分气体
可燃气体浓度
图谱特征提取
卷积运算识别
trace multicomponent gas
flammable gas concentration
map feature extraction
convolution recognition
