两级神经网络的故障诊断方法及其在生化过程中的应用

Double Level Neural Network Method for Fault Diagnosis and Its Application in Biochemical Process

针对具有不能在线检测变量的生化过程 ,提出了一种两级神经网络的故障诊断方法。一级神经网络利用可检测的过程数据建立不可检测量的软测量模型 ;二级神经网络利用软测量模型的估计量进行分析、故障的诊断和分类。最后 ,将这一方法应用到了谷氨酸发酵过程的故障诊断之中 。

A double level neural network method of fault diagnosis is proposed for biochemical process,in which some parameters can not be measured on-line.The first level network is to build soft sensing model of difficult-to-measure process parameters through easy-to-measure parameters.Using these estimates,the second level neural network is to analysis and diagnosis faults.At last,this method is applied to glutamic acid fermentation process,and the results of experiment are listed.