Online Modeling with Semi-Supervised Locally Weighted Partial Least Squares Based on Expectation Maximization Algorithm

doi:10.16182/j.issn1004731x.joss.201801002

Abstract

Abstract: As only small proportion of labeled data can be obtained from chemical processes, an online soft sensing modeling method based on semi-supervised locally weighted partial least squares is proposed . The labeled and unlabeled historical data are accumulated to construct training database. The similarity between the newly measured data and the sample points in the database are calculated and used as the weight of each data point. The semi-supervised locally weighted partial least squares model is constructed, and the Expectation Maximization (EM) algorithm is employed to estimate the parameters of the model. Online prediction achieves simulation results of debutanizer distillation processes, which suggests that the proposed method has good prediction accuracy and stable generalization performance.

Key words: semi-supervised, locally weighted partial least squares, EM algorithm, online modeling

CLC Number:

TP391

Xiong Weili, Xue Mingchen, Li Yanjun. Online Modeling with Semi-Supervised Locally Weighted Partial Least Squares Based on Expectation Maximization Algorithm[J]. Journal of System Simulation, 2018, 30(1): 8-10.

References

[1] 王惠文, 孟洁. 多元线性回归的预测建模方法[J]. 北京航空航天大学学报, 2007, 33(4): 500-504.Wang Huiwen, Meng Jie. Predictive modeling on multivariate linear regression[J]. Journal of Beijing University of Aeronautics and Astronautics, 2007, 33(4): 500-504.
[2] 郑小霞, 钱锋. 基于PCA和最小二乘支持向量机的软测量建模[J]. 系统仿真学报, 2006, 18(3): 739-741. ZHENG Xiaoxia, QIAN Feng. Soft Sensor Modeling Based on PCA and Support Vector Machines[J]. Journal of System Simulation, 2006, 18(3): 739-741.
[3] 王巍, 柴天佑, 赵立杰. 带有稳定学习的递归神经网络动态偏最小二乘建模[J]. 控制理论与应用, 2012, 29(3): 337-341. WANG Wei, CHAI Tian-you, ZHAO Li-jie. Dynamic partial least squares modeling with recurrent neural networks of stable learning[J]. Control Theory & Applications, 2012, 29(3): 337-341.
[4] Shao Weiming, Tian Xueming.Adaptive Soft Sensor for Quality Prediction of Chemical Processes Based on Selective Ensemble of Local Partial Least Squares Models[J]. Chemical Engineering Research and Design (S0263-8762), 2015, 95(3): 113-132.
[5] 王宏伟, 韩云涛, 彭继慎. 基于TSPSO支持向量机红外甲烷传感器动态补偿[J]. 传感技术学报, 2013, 26(9): 1193-1197. WANG Hongwei, HAN Yuntao, PENG Jishen. Dynamic Compensation of Infrared Methane Sensor Based on TSPSO-ε-SVM[J]. CHINESE JOURNAL OF SENSORS AND ACTUATORS, 2013, 26(9): 1193-1197.
[6] 李春富, 郑松, 葛铭. 基于递推非线性部分最小二乘模型的间歇过程批到批优化[J]. 计算机与应用化学, 2011, 28(7): 939-942. Li Chunfu, Zheng Song, Ge Ming. Batch-to-batch optimization of batch processes based on recursive nonlinear partial least squares model[J]. Computers and Applied Chemistry, 2011, 28(7): 939-942.
[7] Cleveland W S.Robust Locally Weighted Regression and Smoothing Scatterplots[J]. Journal of American Statistical Association (S0162-1459), 1979, 74(368): 829-836.
[8] Kim S, Kano M, Nakagawa H, et al.Estimation of Active Pharmaceutical Ingredients Content Using Locally Weighted Partial Least Squares and Statistical Wavelength Selection[J]. International Journal of Pharmaceutics (S0378-5173), 2011, 421(2): 269-274.
[9] 章军, 杨慧中. 基于SVM的苯酚浓度半监督软测量方法[J]. 计算机与应用化学, 2013, 30(20): 1453-1456. Zhang Jun, Yang Huizhong. Semi-supervised soft sensor of Phenol based on SVM[J]. Computers and Applied Chemistry, 2013, 30(20): 1453-1456.
[10] Ge Zhiqiang, Song Zhihuan.Semisupervised Bayesian Method for Soft Sensor Modeling with Unlabeled Data Samples[J]. AIChE Journal (S1547-5905), 2011, 57(8): 2109-2119.
[11] 吴继明. 基于Boosting思想的半监督学习算法研究[D].吉林: 吉林大学, 2014. Wu Jiming. The research of semi-supervised learning based on Boosting[D]. Jilin: Jilin University, 2014.
[12] Kim S, Kano M, Hasebe S, et al.Long-term Industrial Applications of Inferential Control Based on Just-in-time Soft Sensors: Economical Impact and Challenges[J]. Industrial and Engineering Chemistry Research (S0888-5885), 2013, 52(35): 12346-12356.
[13] 唐晓亮, 韩敏. 一种基于极端学习机的半监督学习方法[J]. 大连理工大学学报, 2010, 50(5): 771-776. Tang Xiaoliang, Han Min. A semi-supervised learning method based on extreme learning machine[J]. Journal of Dalian University of Technology, 2010, 50(5): 771-776.
[14] Ge Zhiqiang, Huang Biao, Song Zhihuan.Mixture Semisupervised Principal Component Regression Model and Soft Sensor Application[J]. AIChE Journal (S1547-5905), 2014, 60(2): 533-545.
[15] Ma Ming, Shima K, Huang Biao.A Bayesian Framework for Real-Time Identification of Locally Weighted Partial Least Squares[J]. AIChE Journal (S1547-5905), 2015, 61(2): 518-529.
[16] 熊伟丽, 姚乐, 徐保国. 基于EM算法的青霉素发酵过程多阶段融合建模[J]. 化工学报, 2014, 65(12): 4935-4961. XIONG Weili, YAO Le, XU Baoguo. Multi-stage fusion modeling for penicillin fermentation process based on EM algorithm[J]. CIESC Journal, 2014, 65(12): 4935-4961.
[17] Jin Xing, Huang Biao, Shook D S.Multiple Model LPV Approach to Nonlinear Process Identification with EM Algorithm[J]. Journal of Process Control (S0959-1524), 2011, 21(1): 182-193.
[18] Jin Xing, Huang Biao.Robust Identification of Piecewise/Switching Autoregressive Exogenous Process[J]. AIChE Journal (S1547-5905), 2010, 56(7): 1829-1844.
[19] Yu S P, Yu K, Tresp V, et al.Supervised probabilistic principal component analysis[C]// Proceedings of the 12th ACM International Conference on Knowledge Discovery and Data Mining, Philadelphia, PA, USA:ACM, 2006:464-473.
[20] 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]. CIESC Journal, 2014, 65(11): 4497-4502.