(2.智能过程系统工程教育部工程研究中心, 北京 100029)
(3.中石化炼化工程(集团)股份有限公司, 北京 100101)
【摘要】对于化工过程中带噪声、强耦合的高维数据建模问题,常规的函数连接神经网络(functional link neural networks,FLNN)无法有效地进行处理。为解决该问题,提出一种基于主元分析(principal components analysis,PCA)的函数连接神经网络(PCA-FLNN)。通过对FLNN的函数扩展层进行特征提取,不仅去除变量间的线性相关关系,而且提取数据的主成分,进而简化FLNN学习数据的复杂度。为验证所提方法的有效性,首先采用UCI数据Airfoil Self-Noise对其性能进行验证;随后将所提的方法应用于精对苯二甲酸(purified terephthalic acid,PTA)生产过程建模;与传统FLNN进行对比,标准数据和工业数据的仿真结果表明,PCA-FLNN在处理复杂化工过程数据时具有收敛速度快和建模精度高的特点。
【基金资助】 国家自然科学基金重点基金项目(61533003);国家自然科学基金青年基金项目(61703027); 中央高校基本科研业务费专项资金(ZY1704,JD1708);
Research and application of feature extraction derived functional link neural network
(2.Engineering Research Center of Intelligent Process Systems Engineering, Ministry of Education, Beijing University of Chemical Technology, Beijing, China 100029)
(3.Sinopec Engineering (Group) Co., Ltd., Beijing, China 100101)
【Abstract】Traditional functional link neural network (FLNN) cannot effectively model multi-dimensional, noisy and strongly coupled data in chemical process. A principal component analysis based FLNN (PCA-FLNN) model was proposed to improve modeling effectiveness. Feature extraction of FLNN function extension block not only removed linear correlations between variables but also selected main components of data, which alleviated the complexity of FLNN learning data. The proposed PCA-FLNN model was used to simulate an UCI Airfoil Self-Noise data and purified terephthalic acid (PTA) production process. The simulation results indicate that PCA-FLNN can achieve faster convergence speed with higher modeling accuracy than traditional FLNN.
【Keywords】 functional link neural network; feature extraction; process modeling; purified terephthalic acid;
【Funds】 National Natural Science Foundation of China (61533003, 61703027); Fundamental Research Funds for the Central Universities (ZY1704, JD1708);
 DING S F, SU C Y, YU J Z. An optimizing BP neural network algorithm based on genetic algorithm [J]. Artificial Intelligence Review, 2011, 36 (2): 153–162.
 MOALLEM P, MONADJEMI S A. An efficient MLP-learning algorithm using parallel tangent gradient and improved adaptive learning rates [J]. Connection Science, 2010, 22 (4): 373–392.
 LIU F F, PENG D, HE Y L, et al. Research and chemical application of extreme learning based process neural network [J]. Journal of Shanghai Jiao Tong University, 2014, 48 (7): 977–981 (in Chinese).
 HE Y L, ZHU Q X. A novel robust regression model based on functional link least square (FLLS)and its application to modeling complex chemical processes [J]. Chemical Engineering Science, 2016, 153: 117–128.
 GENG Z Q, WU K Y, HAN Y M. Research and application of FLNN neural network based on AHP [J]. CIESC Journal, 2016, 67 (3): 805–811 (in Chinese).
 MISHRA S K, PANDA G, MEHER S, et al. Exponential functional link artificial neural networks for denoising of image corrupted by Gaussian noise [C] //Proceedings of the 2009 International Conference on Advanced Computer Control. Washington DC, USA: IEEE Computer Society, 2009: 355–359.
 MAJHI R, PANDA G, SAHOO G. Development and performance evaluation of FLANN based model for forecasting of stock markets [J]. Expert Systems with Applications: an International Journal, 2009, 36 (3): 6800–6808.
 HUANG S L, HAO K S, ZHAO W. New improved FLANN approach for dynamic modelling of sensors [J]. International Journal of Computer Applications in Technology, 2011, 41 (1/2): 4–10.
 ZHU B, CHEN Z S, HE Y L, et al. A novel nonlinear functional expansion based PLS (FEPLS) and its soft sensor application [J]. Chemometrics & Intelligent Laboratory Systems, 2017, 161: 108–117.
 HE X Q. Multivariate Statistical Analysis [M]. Beijing: China Renmin University Press, 2012: 135 (in Chinese).
 LEE J M, YOO C K, CHOI S W, et al. Nonlinear process monitoring using kernal principal component analysis [J]. Chem. Eng. Sci., 2004, 59 (1): 223–234
 CHEN Y. Reference–related component analysis: a new method inheriting the advantages of PLS and PCA for separating interesting information and reducing data dimension [J]. Chemometrics and Intelligent Laboratory Systems, 2016, 156: 196–202.
 LU N, YAO Y, GAO F, et al. Two-dimensional dynamic PCA for batch process monitoring [J]. AICh E J., 2005, 51 (12): 3300–3304.
 SONG S O, SHIN D, YOON E S. Analysis of abnormality detection properties of nonlinear PCA methods [J]. IFAC Proceedings Volumes, 2001, 34 (27): 309–314.
 MOORE B. Principal component analysis in linear systems: controllability, observability, and model reduction [J]. IEEE Transactions on Automatic Control, 1981, 26 (1): 17–32.
 JACKSON J E, MUDHOLKAR G S. Control procedures for residuals associated with principal component analysis [J]. Technometrics, 1979, 21 (3): 341–349.
 YUAN Y B, WANGY Y G, CAO F L. Optimization approximation solution for regression problem based on extreme learning machine [J]. Neurocomputing, 2011, 74 (16): 2475–2482.
 LI C F, DAI Y Y, ZHAO J J, et al. Remote sensing monitoring of volcanic ash clouds based on PCA method [J]. Acta Geophysica, 2015: 432–450.
 PENG D Z, ZHANG Y. Dynamics of generalized PCA and MCA learning algorithms [J]. IEEE Transactions on Networks, 2007, 18 (6): 1777–1784.
 ZHANG L, DONG W, ZHANG D, et al. Two-stage image denoising by principal component analysis with local pixel grouping [J]. Pattern Recognition, 2010, 43 (4): 1531–1549.
 LIANG N Y, HUANG G B. A fast and accurate online sequential learning algorithm for feed-forward networks [J]. IEEE Transactions on Networks, 2006, 17 (6): 1411–1423.
 ZHU Z J, ZHANG H W, HAN J, et al. Prediction of coal and gas outburst based on PCA-BP neural network [J]. China Safety Science Journal, 2013, 23 (4): 45–50 (in Chinese).
 HE Y L, WANG X, ZHU Q X. Modeling of acetic acid content in purified terephthalic acid solvent column using principal component analysis based improved extreme learning machine [J]. Control Theory & Applications, 2015, 32 (1): 80–85 (in Chinese).
 SHARIFI R, LANGARI R. Nonlinear sensor fault diagnosis using mixture of probabilistic PCA models [J]. Mechanical Systems & Signal Processing, 2017, 85: 638–650.
 PAO Y H, PHILIPS S M, SOBAJIC D J. Neural-net computing and intelligent control systems [J]. International Journal of Control, 1992, 56 (2): 263–289.
 LÜ J X, WU L N. Nonlinear filtering based on function connected neural network [J]. Journal of Changchun University of Technology, 2006, 27 (4): 305–307 (in Chinese).
 YE S W, SHI Z Z. Neural Networks: a Comprehensive Foundation [M]. Beijing: China Machine Press, 2004: 10–12 (in Chinese).
 HE Y L, XU Y, GENG Z Q, et al. Hybrid robust model integrating with partial least square (IFLNN-PLS) and its application to predicting key process variables [J]. ISA Transactions, 2016, 61: 155–166.
 MOHAMMAD S O. Optimizing functional link neural network learning using modified bee colony on multi-class classifications [M]//JEONG H S, OBAIDAT M, YEN N, PARK J, ed. Advances in Computer Science and Its Applications. Lecture Notes in Electrical Engineering. Berlin: Springer, 2014: 153–159.
 GHAZALI R, BAKAR Z A, HASSIM Y M M, et al. Functional link neural network with modified cuckoo search training algorithm for physical time series forecasting [M]//HUANG D S, BEVILACQUA V, PREMARATNE P, ed. Intelligent Computing Theory. Berlin: Springer, 2014: 285–291.