A novel process monitoring approach based on variational recurrent autoencoder

Abstract Modern manufacturing plants demand not only more intelligent but also safer and more reliable process monitoring systems. With large number of variables measured and stored, significant progress has been made in the past decade on data-driven process monitoring methods and many of them have been successfully applied to monitor various processes. In this work, a new process monitoring method based on variational recurrent autoencoder (VRAE) is proposed. Different from the traditional multivariate statistical process monitoring methods, the proposed method monitors the process in probability space, which enables it to handle process nonlinearity. The proposed method also handles process dynamics through the utilization of recurrent neural network (RNN), which takes the temporal dependency between variables into account. The negative variational score (NVS) is proposed as the monitoring metric. The performance of the proposed method is compared to traditional methods and other neural network based methods on a simple nonlinear simulation and the benchmark simulated Tennessee Eastman process.

[1]  Christos Georgakis,et al.  Disturbance detection and isolation by dynamic principal component analysis , 1995 .

[2]  Ping Zhang,et al.  A comparison study of basic data-driven fault diagnosis and process monitoring methods on the benchmark Tennessee Eastman process , 2012 .

[3]  Leo H. Chiang,et al.  Fault diagnosis in chemical processes using Fisher discriminant analysis, discriminant partial least squares, and principal component analysis , 2000 .

[4]  Jin Wang,et al.  Statistics pattern analysis: A new process monitoring framework and its application to semiconductor batch processes , 2011 .

[5]  Charles M. Bishop.  Variational principal components , 1999 .

[6]  Lakhmi C. Jain,et al.  Recurrent Neural Networks: Design and Applications , 1999 .

[7]  Grigorios Dimitriadis,et al.  Diagnosis of Process Faults in Chemical Systems Using a Local Partial Least Squares Approach , 2008 .

[8]  Takehisa Yairi,et al.  Anomaly Detection Using Autoencoders with Nonlinear Dimensionality Reduction , 2014, MLSDA'14.

[9]  Jeremy S. Smith,et al.  Abnormal Event Detection From Videos Using a Two-Stream Recurrent Variational Autoencoder , 2020, IEEE Transactions on Cognitive and Developmental Systems.

[10]  J. Macgregor,et al.  Monitoring batch processes using multiway principal component analysis , 1994 .

[11]  C. Yoo,et al.  Nonlinear process monitoring using kernel principal component analysis , 2004 .

[12]  Seungjin Choi,et al.  Echo-state conditional variational autoencoder for anomaly detection , 2016, 2016 International Joint Conference on Neural Networks (IJCNN).

[13]  D. R. Baughman,et al.  Neural Networks in Bioprocessing and Chemical Engineering , 1992 .

[14]  Liang Ming,et al.  Abnormal situation management: Challenges and opportunities in the big data era , 2016, Comput. Chem. Eng..

[15]  Geoffrey E. Hinton,et al.  Reducing the Dimensionality of Data with Neural Networks , 2006, Science.

[16]  Christos Georgakis,et al.  Plant-wide control of the Tennessee Eastman problem , 1995 .

[17]  In-Beum Lee,et al.  Fault detection and diagnosis based on modified independent component analysis , 2006 .

[18]  Q. Peter He,et al.  A New Fault Diagnosis Method Using Fault Directions in Fisher Discriminant Analysis , 2005 .

[19]  Carlos F. Alcala,et al.  Reconstruction-based contribution for process monitoring with kernel principal component analysis , 2010, Proceedings of the 2010 American Control Conference.

[20]  Wei Jiang,et al.  Fault diagnosis of rolling bearings with recurrent neural network-based autoencoders. , 2018, ISA transactions.

[21]  Hao Wu,et al.  Deep convolutional neural network model based chemical process fault diagnosis , 2018, Comput. Chem. Eng..

[22]  Jin Wang,et al.  Multivariate Statistical Process Monitoring Based on Statistics Pattern Analysis , 2010 .

[23]  Zhanpeng Zhang,et al.  A deep belief network based fault diagnosis model for complex chemical processes , 2017, Comput. Chem. Eng..

[24]  ChangKyoo Yoo,et al.  Statistical process monitoring with independent component analysis , 2004 .

[25]  David He,et al.  Detection of Pitting in Gears Using a Deep Sparse Autoencoder , 2017 .

[26]  S. Joe Qin,et al.  Statistical process monitoring: basics and beyond , 2003 .

[27]  Margarida Silveira,et al.  Unsupervised Anomaly Detection in Energy Time Series Data Using Variational Recurrent Autoencoders with Attention , 2018, 2018 17th IEEE International Conference on Machine Learning and Applications (ICMLA).

[28]  Mia Hubert,et al.  Overview of PCA-Based Statistical Process-Monitoring Methods for Time-Dependent, High-Dimensional Data , 2015 .

[29]  Weng Khuen Ho,et al.  Neural network systems for multi-dimensional temporal pattern classification , 2005, Comput. Chem. Eng..

[30]  Chong Wang,et al.  Stochastic variational inference , 2012, J. Mach. Learn. Res..

[31]  E. F. Vogel,et al.  A plant-wide industrial process control problem , 1993 .

[32]  In-Beum Lee,et al.  Fault identification for process monitoring using kernel principal component analysis , 2005 .