Robust PPLS model and its applications in process monitoring

doi:10.11949/j.issn.0438-1157.20151439

Abstract

Abstract:

A probability model can be developed by probabilistic partial least squares (PPLS) under the conditions that both principal components and errors satisfy Gaussian distribution. However, the expectation and variance of the Gaussian distribution is susceptible to outliers. As a result, the model is not robust for the real industrial process. This paper improves the robustness of the PPLS model based on the assumption that the raw data satisfy T distribution rather than Gaussian distribution. By adjusting the freedom degree of T distribution, the proposed robust probabilistic partial least squares (RPPLS) model can overcome the shortcomings of PPLS model. Furthermore, on the basis of RPPLS model, two monitoring indicators GT² and GSPE are proposed to monitor the process state and the model changes, respectively. Comparing the monitoring performance in the TE process based on PPLS and RPPLS shows that RPPLS is more effective than PPLS in terms of the fault accuracy and the missing alarm rate.

Key words: robust probabilistic partial least squares algorithm, T distribution, parameter estimation, monitoring indices, model

摘要：

概率偏最小二乘（PPLS）模型建立的条件是主元和误差都服从高斯分布，但是高斯分布的期望和方差容易受到离群点的影响，导致模型的鲁棒性较差。针对PPLS模型的不足，提出一种鲁棒概率偏最小二乘（RPPLS）方法，用拖尾更宽的T分布代替高斯分布，通过调整自由度参数，使模型对含离群点数据的拟合效果更好。更进一步，将RPPLS引入过程监控中，提出GT²和GSPE两个监控指标，分别监控过程的受控状态以及模型关系的变化。PPLS和RPPLS在TE过程监控的应用结果表明RPPLS不仅能更准确检测故障的产生，而且能更有效降低故障的漏报率。

关键词: 鲁棒概率偏最小二乘算法, T分布, 参数估值, 监控指标, 模型

CLC Number:

TP277

CHEN Jiayi, ZHAO Zhonggai, LIU Fei. Robust PPLS model and its applications in process monitoring[J]. CIESC Journal, 2016, 67(7): 2907-2915.

陈家益, 赵忠盖, 刘飞. 鲁棒PPLS模型及其在过程监控中的应用[J]. 化工学报, 2016, 67(7): 2907-2915.

References 24

[1]	KANO M, NAKAGAWA Y. Data-based process monitoring process control and quality improvement: recent developments and applications in steel industry[J]. Computers & Chemical Engineering, 2008, 32(1/2): 12-24. DOI: 10.1016/j.compchemeng.2007.07.005.
[2]	李秀喜, 袁延江. 基于流程模拟的化工故障检测技术[J].化工学报, 2014, 65(11): 4472-4476.DOI: 10.3969/j.issn.0438-1157.2014.11036. LI X X, YUAN Y J. Chemical process fault detection technology based on process simulation[J].CIESC Journal, 2014, 65(11): 4472-4476.DOI: 10.3969/j.issn.0438-1157.2014.11.036.
[3]	SIMOGLOU A, MARTIN E B, MORRIS A J. Multivariate statistical process control of an industrial process fluidised-bed reactor[J]. Control Engineering Practice, 2000, 8(8): 893-909. DOI: 10.1016/S0967-0661(00)00015-0.
[4]	王海清, 宋执环, 王慧. PCA过程监测方法的故障检测行为分析[J].化工学报, 2002, 53(3): 297-301.DOI: 10.3321/j.issn: 0438-1157.2002.03.016. WANG H Q, SONG Z H, WANG H. Fault detection behavior analysis of PCA based process monimring approach[J]. Journal of Chemical Industry and Engineering(China), 2002, 53(3): 297-301.DOI: 10.3321/j.issn: 0438-1157. 2002.03.016.
[5]	WOLD S, SJÖSTRÖMA M, ERIKSSONB L. PLS-regression: a basic tool of chemometrics[J]. Chemometrics and Intelligent Laboratory Systems, 2001, 58(2): 109-130. DOI: 10.1016/S0169- 7439(01)00155-1.
[6]	MEHMOOD T, LILAND K H, SNIPEN L, et al. A review of variable selection methods in partial least squares regression[J]. Chemometrics and Intelligent Laboratory Systems, 2012, 118: 62-69. DOI: 10.1016/j.chemolab.2012.07.010.
[7]	LI S, GAO J, NYAGILO J O, et al. Probabilistic partial least square regression: a robust model for quantitative analysis of Raman spectroscopy data[C]// IEEE International Conference on Bioinformatics and Biomedicine. Atlanta: IEEE, 2011: 526-530.
[8]	LANGE K L, LITTLE R J A, TAYLOR J M G. Robust statistical modeling using the t distribution[J]. Journal of the American Statistical Association, 1989, 84(408): 881-896. DOI: 10.1080/01621459.1989.10478852.
[9]	SHOHAM S. Robust clustering by deterministic agglomeration EM of mixtures of multivariate t-distributions[J]. Pattern Recognition Society, 2002, 35(5): 1127-1142. DOI: 10.1016/S0031- 3203(01)00080-2.
[10]	ARCHAMBEAU C, DELANNAY N, VERLEYSEN M. Robust probabilistic projections[C]//Proceedings of 23rd International Conference on Machine Learning. New York: ACM, 2006: 33-40.
[11]	ARCHAMBEAU C, DELANNAY N, VERLEYSEN M. Mixtures of robust probabilistic principal component analyzers[J]. Neurocomputing, 2008, 71(7/8/9): 1274-1282. DOI: 10.1016/j. neucom.2007.11.029.
[12]	CHEN T, MARTIN E, MONTAGUE G. Robust probabilistic PCA with missing data and contribution analysis for outlier detection[J]. Computational Statistics and Data Analysis, 2009, 53(10): 3706-3716. DOI: 10.1016/j.csda.2009.03.014.
[13]	ZHU J L, GE Z Q, SONG Z H. Robust modeling of mixture probabilistic principal component analysis and process monitoring application[J]. AIChE Journal, 2014, 60(6): 2143-2157. DOI: 10.1002/aic.14419.
[14]	SILVA R, SCHEINES R, GLYMOURl C, et al. Learning the structure of linear latent variable models[J]. Machine Learning Research, 2006, 7(2): 191-246.
[15]	KOURTI T. Application of latent variable methods to process control and multivariate statistical process control in industry[J]. International Journal of Adaptive Control and Signal Processing, 2005, 19(4): 213-246. DOI: 10.1002/acs.859.
[16]	STOICA P, XU L Z, LI J. A new type of parameter estimation algorithm for missing data problems[J]. Statistics & Probability Letters, 2005, 75(3): 219-229. DOI: 10.1016/j.spl.2005.05.018.
[17]	WU C F. On the convergence properties of the EM algorithm[J]. The Annals of Statistics, 1983, 11(1): 95-103.
[18]	MURPHY K P. Machine Learning: A Probabilistic Perspective[M]. London: MIT Press, 2012: 46-49.
[19]	LIU C H, RUBIN D B. ML estimation of the t distribution using EM and its extensions, ECM and ECME[J]. Statistica Sinica, 1995, 5(1): 19-39.
[20]	BISHOP C M. Pattern Recognition and Machine Learning[M]. New York : Springer Press, 2006: 91-93.
[21]	KIM D, LEE I B. Process monitoring based on probabilistic PCA[J]. Chemometrics and Intelligent Laboratory Systems, 2003, 67(2): 109-123. DOI: 10.1016/S0169-7439(03)00063-7.
[22]	马贺贺, 胡益, 侍洪波.基于马氏距离局部离群因子方法的复杂化工过程故障检测[J].化工学报, 2013, 64(5): 1675-1682.DOI: 10.3969/j.issn.0438-1157.2013.05.02. MA H H, HU Y, SHI H B. Fault detection of complex chemical processes using Mahalanobis distance-based local outlier factor[J].CIESC Journal, 2013, 64(5): 1675-1682. DOI: 10.3969/j.issn. 0438-1157.2013.05.02.
[23]	CHIANG L H, RUSSELL E L, BRAATZ R D. Fault Detection and Diagnosis in Industrial Systems[M]. London: Springer London, 2001: 99-106.
[24]	QIN S J, ZHENG Y Y. Quality-relevant and process-relevant fault monitoring with concurrent projection to latent structures[J]. AIChE Journal, 2013, 59(2): 496-504.DOI: 10.1002/aic.13959.