A novel intelligent model integrating PLSR with RBF-Kernel based Extreme Learning Machine: Application to modelling petrochemical process. Issue 1 (2019)

• Record Type: Journal Article
• Authors: Zhu, Qun-Xiong; Zhang, Xiao-Han; Wang, Yanqing; Xu, Yuan; He, Yan-Lin
• Is Part Of: IFAC-PapersOnLine. Volume 52:Issue 1(2019)
• Page Start: 148
• Page End: 153
• Publication Date: 2019
• Subjects: Radial Basis function Kernel -- Partial least squares regression -- Extreme Learning Machine -- Modeling -- High Density Polyethylene process; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2019.06.052
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: With the petrochemical process data getting complicated, building accurate and robust process analysis models has becoming a hot research. In this study, a novel intelligent model integrating PLSR with RBF-Kernel based Extreme Learning Machine (PLSR-RBFKELM) is proposed for overcoming the difficulties found in the conventional extreme learning machine when dealing with collinearity. In the proposed model, radial basis function kernel is employed instead of activation functions in the ELM hidden layer to effectively deal with the high nonlinearity problem of modeling data, and partial least square regression is utilized to solve the collinearity problem. In order to verify the performance, the proposed PLSR-RBFKELM model is applied to modeling one real-world petrochemical process - high density polyethylene process in the steady state. Simulation results demonstrate that the proposed model can achieve good performance in terms of accuracy and stability for static process modeling.