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.

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

Adaptive Kpca Modeling of Nonlinear Systems

SVM Modeling and Application Based on Kernel Function Principal Component Analysis

On Feature Extraction Via Kernels

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

Nonlinear Process Monitoring Using Improved Kernel Principal Component Analysis

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

A Comparative Study of Two Kernel Ideas for Nonlinear Feature Extraction

A method for constructing simplified kernel model based on Kernel-MSE

MODIFIED KERNEL-BASED NONLINEAR FEATURE EXTRACTION

Kernel Mse Algorithm: A Unified Framework for Kfd, Ls-Svm and Krr

Phase Space Reconstruction of Nonlinear Time Series Based on Kernel Method

Nonlinear Component Analysis as a Kernel Eigenvalue Problem

A Kernel Spatial Complexity-Based Nonlinear Unmixing Method of Hyperspectral Imagery.

System Model Identification Based on Kernel Function

Optimizing Kernel Function with Applications to Kernel Principal Analysis and Locality Preserving Projection for Feature Extraction.

Sparse Kernel Extreme Components Analysis

Kernel Multivariate Analysis Framework for Supervised Subspace Learning: A Tutorial on Linear and Kernel Multivariate Methods

The Application of Feature Extraction on Using Kernel Principal Component Analysis Based on Clustering

Improvement of the Kernel Minimum Squared Error Model for Fast Feature Extraction

Feature Extraction Based on Mixture Probabilistic Kernel Principal Component Analysis