Non-Gaussian Process Monitoring Based on NGPP-SVDD

Abstract

A novel NGPP-SVDD statistical process monitor- ing approach is proposed to address non-Gaussian multivariate process systems.Such systems present typical challenges to tra- ditional multivariate process monitoring approaches.The par- ticle swarm optimization (PSO) based NGPP approach is pro- posed to extract the most non-Gaussian component out of the process records and overcome the deficiency of traditional Fas- tICA (fast independent component analysis) algorithm,which is easily trapped in local minimum.Criteria to automatically select the number of non-Gaussian components are also given. Transforming the exacted non-Gaussian components into a fea- ture space using a support vector data description then allows the application of a simple parametric statistical inference.Nu- merical study and industrial melting process application show the efficiency of the proposed method.