Abstract:Kernel principal component analysis (KPCA)-based process monitoring methods have recently shown to be very effective for monitoring nonlinear processes. However, their performances largely depend on the kernel function and currently there is no general rule for kernel selection. Existing methods simply choose the kernel function empirically or experimentally from a given set of candidates. This paper proposes a kernel function learning method for KPCA to learn a kernel function tailored to specific data and explores its potential for KPCA-based process monitoring. Motivated by the manifold learning method maximum variance unfolding (MVU), we obtain the kernel function by optimizing over a family of data-dependent kernels such that the nonlinear structure in input data is unfolded in the kernel feature space and gets more likely to be linear there. Using the optimized kernel, the nonlinear principal components of KPCA which are linear principal components in the kernel feature space can effectively capture the variation in data, and thus the data under normal operating conditions can be more precisely modeled by KPCA for process monitoring. Simulation results on an simple nonlinear system and the benchmark Tennessee Eastman (TE) demonstrate that the optimized kernel functions lead to significant improvement in the performance over the popular Gaussian kernels when used in the KPCA-based process monitoring.

System Model Identification Based on Kernel Function

Novel Nonlinear System Identification Method Based on Kernel Function

Least squares support vector machine based partially linear model identification

Identification methods of nonlinear systems based on the kernel functions

A New Subspace Identification Approach Based on Principal Component Analysis and Noise Estimation

A Spline Kernel-Based Approach for Nonlinear System Identification with Dimensionality Reduction

Adaptive Kpca Modeling of Nonlinear Systems

KPCA Based SVM Identification of Nonlinear System

The nonlinear feature extraction with parsimonious components based on multiple Kernel function

Adaptive Kernel Leaning Networks with Application to Nonlinear System Identification

Nonlinear System Identification Using Scaling Kernel Support Vector Regression

Nonlinear Process Monitoring Using Improved Kernel Principal Component Analysis

Support vector machine based direct inverse-model identification

Learning a Data-Dependent Kernel Function for KPCA-based Nonlinear Process Monitoring

Reproducing wavelet kernel method in nonlinear system identification

One Step Iterative Strategy for Nonlinear System Identification

System identification under non-negativity constraints

SVM Modeling and Application Based on Kernel Function Principal Component Analysis

Nonlinear Mapping and Feature Extraction: KMSE Model and Kernel Principal Component Analysis Technique

Kernel Normalized Mixed-Norm Algorithm For System Identification

Support Vector Regression Based on Scaling Reproducing Kernel for Black-Box System Identification