KERNEL PRINCIPAL COMPONENT ANALYSIS: RADIAL BASIS FUNCTION NEURAL NETWORKS-BASED SOFT-SENSOR MODELING OF POLYMERIZING PROCESS OPTIMIZED BY CULTURAL DIFFERENTIAL EVOLUTION ALGORITHM

• Published in: Instrumentation science & technology Vol. 41; no. 1; pp. 18 - 36
• Main Authors: Wang, Jiesheng, Guo, Qiuping
• Format: Journal Article
• Language: English
• Published: Philadelphia, PA Taylor & Francis Group 01.01.2013; Taylor & Francis; Taylor & Francis Ltd
• Subjects: Algorithmics. Computability. Computer arithmetics; Applied sciences; Artificial intelligence; Computer science; control theory; systems; Connectionism. Neural networks; Control theory. Systems; cultural differential evolution algorithm; Data processing. List processing. Character string processing; Exact sciences and technology; kernel principal component analysis; Memory organisation. Data processing; Modelling and identification; Neural networks; polymerize process; radial basis function neural network; soft sensor; Software; Studies; Theoretical computing; Input output; radial basis function neural network; Evolutionary algorithm; Modeling; Differential evolution; soft sensor; Robustness; System identification; Polyvinyl chloride; Non linear input; Monomer; kernel principal component analysis; Culture; Belief; polymerize process; Dimensionality; cultural differential evolution algorithm; Measurement sensor; Polymerization; Neural network; Real time; Forecasting; Kernel method; Radial basis function; Data fusion; Auxiliary variable; Principal component analysis
• ISSN: 1073-9149; 1525-6030
• DOI: 10.1080/10739149.2012.710884

For forecasting the key technology indicator convention velocity of vinyl chloride monomer (VCM) in the polyvinylchloride (PVC) polymerizing process, a soft-sensor modeling method based on radial basis function neural networks (RBFNN) is proposed. First, a kernel principal component analysis (KPCA) method is adopted to select the auxiliary variables of the soft-sensing model in order to reduce the model dimensionality. Then the structure parameters of the RBFNN are optimized by the cultural differential evolution (CDE) algorithm to realize the nonlinear mapping between input and output variables of the discussed soft-sensor model. In the end, simulation results show that the proposed model can significantly enhance the predictive accuracy and robustness of the technical and economic indexes and satisfy the real-time control requirements of PVC polymerizing production process.