一种数据驱动的工业过程故障诊断与监视方法  

A Data-Driven Method for Industrial Process Fault Diagnosis and Monitoring

作　　者：李元 张安仑 LI Yuan;ZHANG An-lun(College of Information Engineering,Shenyang University of Chemical Technology,Shenyang Liaoning 110142,China)

机构地区：[1]沈阳化工大学信息工程学院,辽宁沈阳110142

出　　处：《计算机仿真》2025年第2期488-493,共6页Computer Simulation

基　　金：国家自然科学基金项目(61673279)。

摘　　要：针对田纳西伊斯曼(Tennessee Eastman, TE)过程数据的预测和监视问题,提出一种基于经验模态分解(Empirical Mode Decomposition, EMD)、主成分分析(Principal component analysis, PCA)和反向传播(Back-propagation, BP)神经网络相结合的方法。首先利用EMD把每一段训练集和测试集数据进行信号分解,得到多个本征模态函数(Intrinsic mode function, IMF)和残差,提取其中能量占比高的成分;再利用PCA技术得到训练集的均值和标准差,将测试集划分成两部分,分别得到其统计量和总统计量;将两部分统计量分别作为神经网络的输入和输出得到误差,最后利用误差的平方和作为控制限进行故障诊断。实验结果表明,数据预测与监视的误差较小,故障检测效果较好。A method based on empirical mode decomposition(EMD),principal component analysis(PCA),and back-propagation(BP)neural networks is proposed to address the nonlinear problem of Tennessee Eastman(TE)process data.Firstly,EMD is used to decompose each training and testing set data into multiple intrinsic mode functions(IMF)and residuals,and extract the components with high energy proportion.Then,the mean and standard deviation of the training set are obtained using PCA technology,and the test set is divided into two parts to obtain its statistical and presidential measures,respectively.The two parts of statistics are used as inputs and outputs of the neural network to obtain errors,and finally,the sum of the squared errors is used as the control limit for fault diagnosis.The experimental results show that the error between data prediction and monitoring is small,and the fault detection effect is good.

关 键 词：田纳西伊斯曼过程 经验模态分解 主成分分析 反向传播神经网络 故障预测与监视 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]