A Just-in-time Learning Soft Sensing Modeling Method Based on Bayesian Gaussian Mixture Model

doi:10.16182/j.issn1004731x.joss.17-0285

Abstract

Abstract: For some time-varying industrial processes with non-Gaussian properties, the model established by the general soft-sensing method is difficult to meet the accuracy requirement. To solve the above problems effectively, a JITL soft sensor modeling method (BGMM) is proposed based on Bayesian Gaussian Mixture Model. For the given training sample set, the number of components of the Gaussian mixture model is optimized by Bayesian Information Criterion (BIC); For new test samples, Gaussian Process Regression (GPR) model is established by using the BGMM similarity criterion for the training samples to find out the most similar set;The model is used to predict the test samples. The effectiveness of the proposed method is verified by modeling and simulating the concentration of butane at the butane tower bottom .

Key words: Gaussian mixture model, just-in-time learning, Bayesian information criterion, Gaussian process regression

CLC Number:

TP273

Qi Cheng, Xiong Weili. A Just-in-time Learning Soft Sensing Modeling Method Based on Bayesian Gaussian Mixture Model[J]. Journal of System Simulation, 2019, 31(8): 1555-1561.

References

[1] Liu J, Chen D S, Shen J F.Development of self-validating soft sensors using fast moving window partial least squares[J]. Industrial & Engineering Chemistry Research (S0888-5885), 2010, 49(22): 11530-11546.
[2] Choi D J, Park H.A hybrid artificial neural network as a software sensor for optimal control of a wastewater treatment process[J]. Water research (S0043-1354), 2001, 35(16): 3959-3967.
[3] Desai K, Badhe Y, Tambe S S, et al.Soft-sensor development for fed-batch bioreactors using support vector regression[J]. Biochemical Engineering Journal (S1369-703X), 2006, 27(3): 225-239.
[4] Ge Zhiqiang.Active probabilistic sample selection for intelligent soft sensing of industrial processes[J]. Chemometrics & Intelligent Laboratory Systems (S0169-7439), 2016, 151: 181-189.
[5] Rasmussen C E.Gaussian processes for machine learning[J]. Lecture Notes in Computer Science (S0129-0657), 2004, 3176: 63-71.
[6] 何志昆, 刘光斌, 赵曦晶, 等. 高斯过程回归方法综述[J]. 控制与决策, 2013, 28(8): 1121-1129.
He Zhikun, Liu Guangbin, Zhao Xijing, et al.Overview of Gaussian process regression[J]. Control and Decision, 2013, 28(8): 1121-1129.
[7] 王春鹏, 于佐军, 孟凡强. 折息移动窗递推PLS算法及其在聚丙烯生产过程中的应用[J]. 化工学报, 2013, 64(12): 4592-4598.
Wang Chunpeng, Yu Zuojun, Meng Fanqiang.Discount moving window recursive PLS algorithm and its application to process of polypropylene production[J]. Journal of Chemical Industry and Engineering, 2013, 64(12): 4592-4598.
[8] Ge Z, Song Z.A comparative study of just-in-time- learning based methods for online soft sensor modeling[J]. Chemometrics and Intelligent Laboratory Systems (S0169-7439), 2010, 104(2): 306-317.
[9] 李庆良, 雷虎民. 一种基于即时学习的多模型在线建模方法[J]. 四川大学学报(工程科学版), 2010, 42(1): 196-200.
Li Qingliang, Lei Humin.An Online Multiple-model Modeling Method Based on Lazy Learning[J]. Journal of Sichuan University(Engineering Science Edition), 2010, 42(1): 196-200.
[10] Witten I H, Frank E, Hall M A, et al.Data Mining: Practical machine learning tools and techniques[J]. BioMedical Engineering Online (S1475-925X), 2006, 5(1).
[11] 李昌利, 沈玉利. 期望最大算法及其应用(1)[J]. 计算机工程与应用, 2008, 44(29): 61-64.
Li Changli, Shen Yuli.Tutorial of EM algorithm and its application(1)[J]. Computer Engineering and Applications, 2008, 44(29): 61-64.
[12] Shimo-onoda K, Tanaka T, Furushima K, et al. Akaike's information criterion for a measure of linkage disequilibrium[J]. Journal of human genetics (S1434-5161), 2002, 47(12): 0649-0655.
[13] Zanini A, Woodbury A D.Contaminant source reconstruction by empirical Bayes and Akaike's Bayesian Information Criterion[J]. Journal of contaminant hydrology, 2016, 185: 74-86.
[14] Mehrjou A, Hosseini R, Araabi B N.Improved Bayesian information criterion for mixture model selection[J]. Pattern Recognition Letters (S0167-8655), 2016, 69: 22-27.
[15] Han J, Zhang X P, Wang F.Gaussian Process Regression Stochastic Volatility Model for Financial Time Series[J]. IEEE Journal of Selected Topics in Signal Processing (S1941-0484), 2016, 10(6): 1015-1028.
[16] 阮宏镁, 田学民, 王平. 基于联合互信息的动态软测量方法[J]. 化工学报, 2014, 65(11): 4497-4502.
Ruan Hongmei, Tian Xuemin, Wang Ping.Dynamic soft sensor method based on joint mutual information[J]. Journal of Chemical Industry and Engineering, 2014, 65(11): 4497-4502.