Performance monitoring of non-gaussian chemical processes with modes-switching using globality-locality preserving projection

Xin Peng, Yang Tang, Wenli Du, Feng Qian

PDF(531 KB)
PDF(531 KB)
Front. Chem. Sci. Eng. ›› 2017, Vol. 11 ›› Issue (3) : 429-439. DOI: 10.1007/s11705-017-1675-6
RESEARCH ARTICLE
RESEARCH ARTICLE

Performance monitoring of non-gaussian chemical processes with modes-switching using globality-locality preserving projection

Author information +
History +

Abstract

In this paper, we propose a novel performance monitoring and fault detection method, which is based on modified structure analysis and globality and locality preserving (MSAGL) projection, for non-Gaussian processes with multiple operation conditions. By using locality preserving projection to analyze the embedding geometrical manifold and extracting the non-Gaussian features by independent component analysis, MSAGL preserves both the global and local structures of the data simultaneously. Furthermore, the tradeoff parameter of MSAGL is tuned adaptively in order to find the projection direction optimal for revealing the hidden structural information. The validity and effectiveness of this approach are illustrated by applying the proposed technique to the Tennessee Eastman process simulation under multiple operation conditions. The results demonstrate the advantages of the proposed method over conventional eigendecomposition-based monitoring methods.

Keywords

non-Gaussian processes / subspace projection / independent component analysis / locality preserving projection / finite mixture model

Cite this article

Download citation ▾
Xin Peng, Yang Tang, Wenli Du, Feng Qian. Performance monitoring of non-gaussian chemical processes with modes-switching using globality-locality preserving projection. Front. Chem. Sci. Eng., 2017, 11(3): 429‒439 https://doi.org/10.1007/s11705-017-1675-6

References

[1]
Yin S, Shi  P, Yang H . Adaptive fuzzy control of strict-feedback nonlinear time-delay systems with unmodeled dynamics. IEEE Transactions on Cybernetics, 2016, 46(8): 1926–1938
CrossRef Google scholar
[2]
Yin S, Zhu  X P, Qiu  J B, Gao  H J. State estimation in nonlinear system using sequential evolutionary filter. IEEE Transactions on Industrial Electronics, 2016, 63(6): 3786–3794
CrossRef Google scholar
[3]
Yin S, Gao  H, Qiu J ,  Kaynak O . Descriptor reduced-order sliding mode observers design for switched systems with sensor and actuator faults. Automatica, 2017, 76: 282–292
CrossRef Google scholar
[4]
Tong C D, Shi  X H. Decentralized monitoring of dynamic processes based on dynamic feature selection and informative fault pattern dissimilarity. IEEE Transactions on Industrial Electronics, 2016, 63(6): 3804–3814
CrossRef Google scholar
[5]
Stubbs S, Zhang  J, Morris J . Fault detection in dynamic processes using a simplified monitoring-specific CVA state space modelling approach. Computers & Chemical Engineering, 2012, 41: 77–87
CrossRef Google scholar
[6]
Nomikos P, MacGregor  J F. Monitoring batch processes using multiway principal component analysis. AIChE Journal, 1994, 40(8): 1361–1375
CrossRef Google scholar
[7]
Xiao Z B, Wang  H G, Zhou  J W. Robust dynamic process monitoring based on sparse representation preserving embedding. Journal of Process Control, 2016, 40: 119–133
CrossRef Google scholar
[8]
Qin S J. Statistical process monitoring: Basics and beyond. Journal of Chemometrics, 2003, 17(8-9): 480–502
CrossRef Google scholar
[9]
Ge Z Q, Song  Z H, Gao  F R. Review of recent research on data-based process monitoring. Industrial & Engineering Chemistry Research, 2013, 52(10): 3543–3562
CrossRef Google scholar
[10]
Dong D, McAvoy  T J. Nonlinear principal component analysis — based on principal curves and neural networks. Computers & Chemical Engineering, 1996, 20(1): 65–78
CrossRef Google scholar
[11]
Antory D, Irwin  G W, Kruger  U, McCullough G . Improved process monitoring using nonlinear principal component models. International Journal of Intelligent Systems, 2008, 23(5): 520–544
CrossRef Google scholar
[12]
Silva R G. Condition monitoring of the cutting process using a self-organizing spiking neural network map. Journal of Intelligent Manufacturing, 2010, 21(6): 823–829
CrossRef Google scholar
[13]
Wang B, Yan  X F, Jiang  Q C. Independent component analysis model utilizing de-mixing information for improved non-Gaussian process monitoring. Computers & Industrial Engineering, 2016, 94: 188–200
CrossRef Google scholar
[14]
Ge Z, Song  Z. Process monitoring based on independent component analysis-principal component analysis (ICA-PCA) and similarity factors. Industrial & Engineering Chemistry Research, 2007, 46(7): 2054–2063
CrossRef Google scholar
[15]
Choi S W, Lee  I B. Nonlinear dynamic process monitoring based on dynamic kernel PCA. Chemical Engineering Science, 2004, 59(24): 5897–5908
CrossRef Google scholar
[16]
Zhang Y W, An  J Y, Zhang  H L. Monitoring of time-varying processes using kernel independent component analysis. Chemical Engineering Science, 2013, 88: 23–32
CrossRef Google scholar
[17]
Kano M, Tanaka  S, Hasebe S ,  Hashimoto I ,  Ohno H. Monitoring independent components for fault detection. AIChE Journal, 2003, 49(4): 969–976
CrossRef Google scholar
[18]
Zhang Y, Zhang  Y. Fault detection of non-Gaussian processes based on modified independent component analysis. Chemical Engineering Science, 2010, 65(16): 4630–4639
CrossRef Google scholar
[19]
Ge Z Q, Xie  L, Kruger U ,  Song Z H . Local ICA for multivariate statistical fault diagnosis in systems with unknown signal and error distributions. AIChE Journal, 2012, 58(8): 2357–2372
CrossRef Google scholar
[20]
Hsu C C, Chen  M C, Chen  L S. A novel process monitoring approach with dynamic independent component analysis. Control Engineering Practice, 2010, 18(3): 242–253
CrossRef Google scholar
[21]
Rashid M M, Yu  J. A new dissimilarity method integrating multidimensional mutual information and independent component analysis for non-Gaussian dynamic process monitoring. Chemometrics and Intelligent Laboratory Systems, 2012, 115: 44–58
CrossRef Google scholar
[22]
Costa J, Hero  A O. Geodesic entropic graphs for dimension and entropy estimation in manifold learning. Signal Processing. IEEE Transactions on, 2004, 52(8): 2210–2221
CrossRef Google scholar
[23]
Lin T, Zha  H. Riemannian manifold learning. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008, 30(5): 796–809
CrossRef Google scholar
[24]
Tenenbaum J B ,  de Silva V ,  Langford J C . A global geometric framework for nonlinear dimensionality reduction. Science, 2000, 290(5500): 2319–2323
CrossRef Google scholar
[25]
Roweis S T, Saul  L K. Nonlinear dimensionality reduction by locally linear embedding. Science, 2000, 290(5500): 2323–2326
CrossRef Google scholar
[26]
Zhang Z Y, Zha  H Y. Principal manifolds and nonlinear dimensionality reduction via tangent space alignment. SIAM Journal on Scientific Computing, 2004, 26(1): 313–338
CrossRef Google scholar
[27]
He X, Niyogi  P. Locality preserving projections. In: Proceedings of the Neural Information Processing Systems. Neural Information Processing Systems Foundation. Cambridge: MIT Press, 2004, 153
[28]
Luo L. Process monitoring with global-local preserving projections. Industrial & Engineering Chemistry Research, 2014, 53(18): 7696–7705
CrossRef Google scholar
[29]
Yu J, Qin  S J. Multimode process monitoring with Bayesian inference-based finite Gaussian mixture models. AIChE Journal, 2008, 54(7): 1811–1829
CrossRef Google scholar
[30]
Fan M, Ge  Z Q, Song  Z H. Adaptive Gaussian mixture model-based relevant sample selection for JITL soft sensor development. Industrial & Engineering Chemistry Research, 2014, 53(51): 19979–19986
CrossRef Google scholar
[31]
Yu J. A nonlinear kernel Gaussian mixture model based inferential monitoring approach for fault detection and diagnosis of chemical processes. Chemical Engineering Science, 2012, 68(1): 506–519
CrossRef Google scholar
[32]
Wen Q, Ge  Z, Song Z . Data-based linear Gaussian state-space model for dynamic process monitoring. AIChE Journal, 2012, 58(12): 3763–3776
CrossRef Google scholar
[33]
Ge Z, Kruger  U, Lamont L ,  Xie L, Song  Z. Fault detection in non-Gaussian vibration systems using dynamic statistical-based approaches. Mechanical Systems and Signal Processing, 2010, 24(8): 2972–2984
CrossRef Google scholar
[34]
Hyvarinen A, Oja  E. Independent component analysis: Algorithms and applications. Neural Networks, 2000, 13(4-5): 411–430
CrossRef Google scholar
[35]
Zhang M G, Ge  Z Q, Song  Z H, Fu  R W. Global-local structure analysis model and its application for fault detection and identification. Industrial & Engineering Chemistry Research, 2011, 50(11): 6837–6848
CrossRef Google scholar
[36]
Figueiredo M A T ,  Jain A K . Unsupervised learning of finite mixture models. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 24(3): 381–396
CrossRef Google scholar
[37]
Te-Won L, Lewicki  M S, Sejnowski  T J. ICA mixture models for unsupervised classification of non-Gaussian classes and automatic context switching in blind signal separation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(10): 1078–1089
CrossRef Google scholar
[38]
Downs J J, Vogel  E F. A plant-wide industrial process control problem. Computers & Chemical Engineering, 1993, 17(3): 245–255
CrossRef Google scholar
[39]
Lee J M, Qin  S J, Lee  I B. Fault detection and diagnosis based on modified independent component analysis. AIChE Journal, 2006, 52(10): 3501–3514
CrossRef Google scholar

Acknowledgement

This work was supported by the National Natural Science Foundation of China (Grant Nos. 61333010, 61590923 and 21376077).

RIGHTS & PERMISSIONS

2017 Higher Education Press and Springer-Verlag Berlin Heidelberg
AI Summary AI Mindmap
PDF(531 KB)

Accesses

Citations

Detail

Sections
Recommended

/