摘要
针对分离过程 ,从整体上提出了软测量软件开发策略以及开发过程中要注意的问题。首先采用时序MT—NT法和神经网络方法对数据进行过失误差侦破和数据校正的预处理 ,利用数据相关性和主元分析相结合的方法进行辅助变量选择 ,然后利用神经网络的智能性来建立软测量模型。利用以上算法同时采用面向对象的编程技术 ,开发出了对分离过程具有一定通用性的软测量软件。软件的工业实例应用表明 ,该软件可有效的预测过程变量 。
The article brings forward a development strategy about soft-sensing software aiming at separation process on the whole and the important questions which should be noticed in the development software process.Firstly,modified time series method and artificial neural network method are integrated to deal with the process data.Gross error is detected and the data is rectified.Then PCA method and Data relativity analysis method are used to ascertain the secondary variable.Finally the intelligent of ANN technology as an effective method is employed in this software to get soft-sensing algorithm modules.With all the algorithms above mentioned and the object oriented programming technology,the soft-sensing software used in separation process with certain intelligent property and adaptability is developed.The software is used in many industry examples.The results show that this software can get exact primary variable results and the results can meet the industrial demand.
出处
《化工时刊》
CAS
2004年第4期4-7,共4页
Chemical Industry Times
基金
山东省教育厅资助项目 :Jolco6
