基于改进LSTM神经网络的化工过程故障诊断

摘要/Abstract

摘要： 针对现代化工过程中数据非线性、高维度以及动态时序等特点,传统的故障诊断模型对化工过程的故障诊断精度较低.基于此,设计了一种基于改进的长短时记忆神经网络(LSTM)故障诊断方法.首先,将采集的故障数据输入卷积神经网络(CNN),对数据进行特征提取和降维;其次,将处理过的数据输入改进的LSTM网络,进行深层特征提取;最后,把提取的深层特征信息输入到注意力机制进行特征“聚焦”,实现特征融合后输入softmax分类器实现故障分类.由田纳西-伊斯曼(TE)过程诊断实验结果表明,基于改进的LSTM网络的故障诊断方法在故障分类精度、训练速度方面都更优于递归神经网络(RNN)、门控循环神经网络(GRU)、卷积神经网络(CNN)和深度自编码网络(DAEN),在实际化工过程的应用有一定的优势.

关键词: 化工过程, 深度学习, 注意力机制, 故障诊断

Abstract: According to the characteristics of data nonlinearity, high dimension, and dynamic time sequence in modern chemical processes, traditional methods have low accuracy for fault diagnosis. Therefore, a fault diagnosis method based on improved long short-term memory (LSTM) neural network is designed in this paper. First, the collected fault data were input into a convolutional neural network (CNN) to extract the features and reduce the dimension of the data. Secondly, the processed data was input into the improved LSTM network for deep feature extraction. Finally, the extracted deep feature information was input into the attention mechanism for feature “focusing” to realize feature fusion, and then input into the softmax classifier to realize fault classification. Simulation results on the Tennessee-Eastman (TE) dataset demonstrate that the proposed method is better than recurrent neural network (RNN), gated recurrent unit (GRU), CNN, and deep auto-encoder network (DAEN) in fault classification accuracy and diagnose speed, which has certain advantages in the application of the practical chemical process.

Key words: chemical process, deep learning, attention mechanism, fault diagnosis

中图分类号:

TP277

杜先君, 邱小彧. 基于改进LSTM神经网络的化工过程故障诊断[J]. 兰州理工大学学报, 2023, 49(6): 72-79.

DU Xian-jun, QIU Xiao-yu. Fault diagnosis for chemical process based on an improved LSTM neural network[J]. Journal of Lanzhou University of Technology, 2023, 49(6): 72-79.

参考文献

[1] 蒋浩,陈丙珍.化工过程稳定性分析研究进展[J].化工学报,2018,69(1):76-87.
[2] 戴一阳,赵劲松,陈丙珍.化工过程混合故障诊断系统的应用[J].化工学报,2010,61(2):342-346.
[3] 姚羽曼,罗文嘉,戴一阳.数据驱动方法在化工过程故障诊断中的研究进展[J].化工进展,2010,40(4):1755-1764.
[4] 梁北辰,戴景民.偏最小二乘法在系统故障诊断中的应用[J].哈尔滨工业大学学报,2020,52(3):162-170.
[5] 李元良,丁坤,陈富东,等.基于快速过采样主成分分析法的光伏阵列故障诊断[J].电网技术,2019,43(1):308-315.
[6] 陈鹏,赵小强,朱奇先.基于VMD-MPE-KPCA特征提取与MRVM相混合的滚动轴承故障诊断方法[J].兰州理工大学学报,2020,46(5):92-99.
[7] 孔祥玉,解建,罗家宇,等.基于改进高效偏最小二乘的质量相关故障诊断[J].控制理论与应用,2020,37(12):2645-2653.
[8] 沈欧,李元,吕锋.基于PCA故障诊断的化工过程多智能体优化策略[J].吉林大学学报(工学版),2004,34(第15届中国过程控制会议专辑):203-206.
[9] KINI K R,MADAKYARU M.Improved process monitoring strategy using kantorovich distance-independent component analysis:an application to Tennessee Eastman process[J].IEEE Access,2020,8:205863-205877.
[10] DU B,KONG X,FENG X.Generalized principal component analysis-based subspace decomposition of fault deviations and its application to fault reconstruction[J].IEEE Access,2020,8:34177-34186.
[11] ZHANG C,GUO Q,LI Y.Fault detection in the Tennessee Eastman benchmark process using principal component difference based on K-nearest neighbors[J].IEEE Access,2020,8:49999-50009.
[12] JING C,HOU J.SVM and PCA based fault classification approaches for complicated industrial process[J].Neurocomputing,2015,167:636-642.
[13] CHENG Y J,LIU H M,CHEN Y,et al.Fault diagnosis for rolling bearing based on SIFT-KPCA and SVM[J].Engineering Computations:International Journal for Computer-aided Engineering and Software,2017,34(1):53-65.
[14] YANG G,GU X.Fault diagnosis of complex chemical processes based on enhanced Naive Bayesian method[J].IEEE Transactions on Instrumentation and Measurement,2020,69(7):4649-4658.
[15] ZHAO S,CHENG X,LI J.Sensor fault diagnosis based on adaptive arc fuzzy DBN-petri net[J].IEEE Access,2021,9:20305-20317.
[16] ZHANG X,HUANG T,WU B,et al.Multi-model ensemble deep learning method for intelligent fault diagnosis with high-dimensional samples[J].Frontiers of Mechanical Engineering,2021,16(2):340-352.
[17] QIN F W,BAI J,YUAN W Q,et al.Research on intelligent fault diagnosis of mechanical equipment based on sparse deep neural networks[J].Journal of Vibroengineering,2017,19(4):2439-2455.
[18] ZHANG Z,JIANG W,GENG J,et al.Fault diagnosis based on non-negative sparse constrained deep neural networks and Dempster-Shafer theory[J].IEEE Access,2020,8:18182-18195.
[19] HAN S,YANG F,YANG G,et al.Electrical equipment identification in infrared images based on ROI-selected CNN method[J].Electric Power Systems Research,2020,188:106534.
[20] CHEN Z,MAURICIO A,LI W,et al.A deep learning method for bearing fault diagnosis based on cyclic spectral coherence and convolutional neural networks[J].Mechanical Systems and Signal Processing,2020,140:106683.
[21] XU J,LIU X,WANG B,et al.Deep belief network-based gas path fault diagnosis for turbofan engines[J].IEEE Access,2019,7:170333-170342.
[22] WU H,ZHAO J.Deep convolutional neural network model based chemical process fault diagnosis[J].Computers & Chemical Engineering,2018,115:185-197.
[23] HAN Y,DING N,GENG Z,et al.An optimized long short-term memory network based fault diagnosis model for chemical processes[J].Journal of Process Control,2020,92:161-168.
[24] 陈保家,陈学力,沈保明,等.CNN-LSTM深度神经网络在滚动轴承故障诊断中的应用[J].西安交通大学学报,2021,55(6):28-36.
[25] 邢砾文,姚文凯,黄莹.基于深度学习的含未知复合故障多传感器信号故障诊断[J].重庆大学学报(自然科学版),2020,43(9):93-100.
[26] 郑一珍,牛蔺楷,熊晓燕,等.基于一维卷积神经网络的圆柱滚子轴承保持架故障诊断[J].振动与冲击,2021,40(19):230-238.
[27] GAO X,YANG F,FENG E.A process fault diagnosis method using multi-time scale dynamic feature extraction based on convolutional neural network[J].The Canadian Journal of Chemical Engineering,2020,98(6):1280-1292.
[28] LI W,LIU Y,LI Y,et al.Series arc fault diagnosis and line selection method based on recurrent neural network[J].IEEE Access,2020,8:177815-177822.
[29] LEI J,LIU C,JIANG D.Fault diagnosis of wind turbine based on long short-term memory networks[J].Renewable Energy,2019,133:422-432.
[30] ZHANG H,SUN S,BO J.Sequential fault diagnosis based on LSTM neural network[J].IEEE Access,2018,6(99):12929-12939.
[31] 王太勇,王廷虎,王鹏,等.基于注意力机制BiLSTM的设备智能故障诊断方法[J].天津大学学报,2020,53(6):601-608.
[32] 张立鹏,毕凤荣,程建刚,等.基于注意力BiGRU的机械故障诊断方法研究[J].振动与冲击,2021,40(5):113-118.
[33] MU K,LUO L,WANG Q,et al.Industrial process monitoring and fault diagnosis based on temporal attention augmented deep network[J].Journal of Information Processing Systems,2021,17(2):242-252.
[34] YIN S,DING S X,HAGHANI A,et al.A comparison study of basic data-driven fault diagnosis and process monitoring methods on the benchmark Tennessee Eastman process-science direct[J].Journal of Process Control,2012,22(9):1567-1581.