Identification of linear parameter varying systems with variational Bayesian algorithm

doi:10.11949/j.issn.0438-1157.20171563

Abstract

Abstract:

Linear parameter varying (LPV) method is an effective tool for nonlinear process modeling, which converts modeling of multi-stage nonlinear complex process into identification of multiple linear models. Various disturbance factors in industrial processes result in stochasticity of system modeling and uncertainty of model parameters. Identification of LPV models was studied under the variational Bayesian (VB) framework. After prior probability distributions were assigned to variables and parameters, posterior distributions of these variables and parameters were estimated by maximizing lower limits of objective functions. This full Bayesian system identification approach not only provided point estimates of parameters, but also quantified uncertainty of estimation. Numerical simulation on typical two-stage process and continuous stirred tank reactor (CSTR) demonstrated effectiveness and superiority of the proposed method.

Key words: nonlinear process, linear parameter varying system, multiple models, variational Bayesian algorithm, parameter estimation

摘要：

线性变参数系统（LPV）将多阶段、非线性的过程建模转化为线性多模型的辨识问题，是解决非线性过程建模的一个有效手段。由于实际工业过程存在各种干扰因素，导致被建模系统呈现随机性及模型参数的不确定性。针对这一问题，考虑采用变分贝叶斯（VB）算法对LPV模型进行辨识。该算法首先给定参数相应的先验分布，通过最大化目标函数的下界，从而估计得到参数的后验分布。不仅可实现对参数的点估计，同时量化了估计值的不确定性。针对典型二阶过程和连续搅拌反应釜（CSTR），运用提出的算法进行仿真实验，表明了该贝叶斯估计方法的优越性。

关键词: 非线性过程, 线性变参数系统, 多模型, 变分贝叶斯算法, 参数估计

CLC Number:

TP301.6

LI Hanshuang, ZHAO Zhonggai, LIU Fei. Identification of linear parameter varying systems with variational Bayesian algorithm[J]. CIESC Journal, 2018, 69(7): 3125-3134.

李寒霜, 赵忠盖, 刘飞. 基于变分贝叶斯算法的线性变参数系统辨识[J]. 化工学报, 2018, 69(7): 3125-3134.

References 30

[1]	VERDULT V, VERHAEGEN M. Subspace identification of multivariable linear parameter-varying systems[J]. Automatica, 2002, 38(5):805-814.
[2]	TOTH R, HEUBERGER P S C, HOF P M J V D. Asymptotically optimal orthonormal basis functions for LPV system identification[J]. Automatica, 2009, 45(6):1359-1370.
[3]	ZHAO Y, HUANG B, SU H Y, et al. Prediction error method for identification of LPV models[J]. Journal of Process Control, 2012, 22(1):180-193.
[4]	TEHRANI E S, KEARNEY R E. A non-parametric linear parameter varying approach for identification of linear time-varying systems[J]. IFAC-Papers Online, 2015, 48(28):733-738.
[5]	HOFFMANN C, WERNER H. A survey of linear parameter-varying control applications validated by experiments or high-fidelity simulations[J]. IEEE Transactions on Control Systems Technology, 2015, 23(2):416-433.
[6]	ZHAO P, NAGAMUNE R. Switching LPV controller design under uncertain scheduling parameters[J]. Automatica, 2017, 76:243-250.
[7]	SHAMMA J S, ATHANS M. Guaranteed properties of gain scheduled control for linear parameter-varying plants[J]. Automatica, 1991, 27(3):559-564.
[8]	ZHU Y C, XU Z H. A method of LPV model identification for control[J]. IFAC Proceedings Volumes, 2008, 17(1):5018-5023.
[9]	XU Z H, ZHAO J, QIAN J X, et al. Nonlinear MPC using an identified LPV model[J]. Industrial & Engineering Chemistry Research, 2009, 48(6):3043-3051.
[10]	JIN X, HUANG B, SHOOK D S. Multiple model LPV approach to nonlinear process identification with EM algorithm[J]. Journal of Process Control, 2011, 21(1):182-193.
[11]	YANG X Q, HUANG B, GAO H J. A direct maximum likelihood optimization approach to identification of LPV time-delay systems[J]. Journal of the Franklin Institute, 2016, 353(8):1862-1881.
[12]	GOODFELLOW L, BENGIO Y, COURVILLE A. Deep Learning[M]. Massachusetts:The MIT Press, 2016:54-57.
[13]	ATTIAS H. A variational Bayesian framework for graphical models[J]. International Conference on Neural Information Processing Systems, 1999, 12:209-215.
[14]	YANG X Q, YIN S. Variational Bayesian inference for FIR models with randomly missing measurements[J]. IEEE Transactions on Industrial Electronics, 2016, 64(5):1-9.
[15]	ZHAO Y J, FATEHI A, HUANG B. Robust estimation of ARX models with time varying time delays using variational Bayesian approach[J]. IEEE Transactions on Cybernetics, 2017, 99:1-11.
[16]	MA Z, RANA P K, TAGHIA J, et al. Bayesian estimation of Dirichlet mixture model with variational inference[J]. Pattern Recognition, 2014, 47(9):3143-3157.
[17]	ALA-LUHTALA J, SARKKA S, PICHE R. Gaussian filtering and variational approximations for Bayesian smoothing in continuousdiscrete stochastic dynamic systems[J]. Signal Processing, 2015, 111:124-136.
[18]	BACHNAS A A, TOTH R, LUDLAGE J H A, et al. A review on datadriven linear parameter-varying modeling approaches:a high-purity distillation column case study[J]. Journal of Process Control, 2014, 24(4):272-285.
[19]	LU Y J, KHATIBISEPEHR S, HUANG B. A variational Bayesian approach to identification of switched ARX models[J]. IEEE Conference on Decision and Control, 2014, 2015:2542-2547.
[20]	HUANG J Y, JI G L, ZHU Y C, et al. Identification of multi-model LPV models with two scheduling variables[J]. Journal of Process Control, 2012, 22(7):1198-1208.
[21]	YOU J, LU J G, ZHU Y C, et al. Identification of multimodel LPV models with asymmetric Gaussian weighting function[J]. Journal of Applied Mathematics, 2013, 2013(2):1-12.
[22]	YOU J, YANG Q M, LU J G, et al. Identification of LPV models with non-uniformly spaced operating points by using asymmetric Gaussian weights[J]. Chinese Journal of Chemical Engineering, 2014, 22(6):795-798.
[23]	BERNARDO J M, BAYARRI M J, BERGER J O, et al. The variational Bayesian EM algorithm for incomplete data:with application to scoring graphical model structures[J]. Bayesian Statistics, 2003, 78(78):417-429.
[24]	LU Y J, HUANG B, KHATIBISEPEHR S. A variational Bayesian approach to robust identification of switched ARX Models[J]. IEEE Transactions on Cybernetics, 2017, 46(12):3195-3208.
[25]	LU Y J, HUANG B. Robust multiple-model LPV approach to nonlinear process identification using mixture t distributions[J]. Journal of Process Control, 2014, 24(9):1472-1488.
[26]	TAKEKAWA T, FUKAI T. A novel view of the variational Bayesian clustering[J]. Neurocomputing, 2009, 72(13/14/15):3366-3369.
[27]	CHEN L, TULSYAN A, HUANG B, et al. Multiple model approach to nonlinear system identification with an uncertain scheduling variable using EM algorithm[J]. Journal of Process Control, 2013, 23(10):1480-1496.
[28]	DENG J, HUANG B. Identification of nonlinear parameter varying systems with missing output data[J]. AIChE Journal, 2012, 58(11):3454-3467.
[29]	XIONG W L, YANG X Q, HUANG B, et al. Multiple-model based linear parameter varying time-delay system identification with missing output data using an expectation-maximization algorithm[J]. Industrial & Engineering Chemistry Research, 2014, 53(27):11074-11083.
[30]	YANG X Q, YIN S. Robust global identification and output estimation for LPV dual-rate systems subjected to random output timedelays[J]. IEEE Transactions on Industrial Informatics, 2017, 13(6):2876-2885.