Machine Learning Kernel Methods for Statistical Process Monitoring: A Review and Case Studies

This article proposes the application of machine learning kernel methods in statistical process monitoring. It addresses the limitations of traditional statistical process monitoring methods in handling nonlinear and high-dimensional data, and presents a kernel-based monitoring framework.

The method is validated through two case studies. The first case study demonstrates the method's effectiveness in monitoring a chemical reaction process. The authors used kernel principal component analysis (KPCA) to reduce the dimensionality of the reaction process and a kernel support vector machine (SVM)-based anomaly detection method to monitor the process. The experimental results show that the method can effectively detect anomalies in the reaction process.

The second case study validates the method's application in monitoring a machine tool process. The authors used a kernel principal component analysis-based and a kernel support vector machine-based monitoring method to monitor the quality control of the machine tool process. Experimental results demonstrate that the method can effectively detect anomalies in the machine tool process and accurately classify them.

In conclusion, this article presents the application of machine learning kernel methods in statistical process monitoring, and validates the method's effectiveness and practicality through two case studies.