摘要
研究了一种应用于精馏塔的多源信息融合的故障诊断方法。选择影响精馏塔系统生产质量的主要参数底温、顶温等作为融合对象。首先对参数进行归一化处理,并利用主元分析法对提取的底温、顶温等特征数据进行处理,这样既降低了输入的维数,同时又提高了输入参量特征的相互独立性。然后通过神经网络对主元分析法处理后的特征矢量进行推理分类,得到精馏塔的故障诊断结果。以在一定压力下,间接反映氯乙烯精馏塔产品浓度的塔板温度为诊断对象,仿真结果表明该方法具有很好的诊断效果。
A method of fault diagnosis basing on multi-source information fusion is applied in the distillation column. Extracting the main parameters that affects the production quality of the distillation column system such as temperature at the end, the top temperature and so on as a fusion of object and data preprocessing and feature extraction by using princi pal component analysis, through linearly associating the points in higher dimension space and projecting them to lower di mension space, not only reducing the input dimension, but also enhancing the mutual independence of the characteristics of the input parameters. Then, we can classify the data by neural network. Under a certain pressure, plate temperature that indirectly reflect the concentration of vinyl chloride distillation column as a diagnostic object , the simulation results show that the method is a good diagnostic result.
出处
《武汉理工大学学报》
CAS
CSCD
北大核心
2012年第11期135-138,共4页
Journal of Wuhan University of Technology
基金
湖北省自然科学基金(2010CDB11101)
关键词
多源信息
故障诊断
主元分析
神经网络
multi-source information
fault diagnosis
principal component analysis
neural network
