Weighted Semi-supervised Orthogonal Factor Analysis Model for Quality-Related Process Monitoring

Probabilistic model has already been widely used for process monitoring. However, the obtained factors may contain quality-unrelated information, which is harmful to the quality-related process monitoring. Meanwhile, considering the situation of unequal sample rates of process and quality variables, a semi -supervised orthogonal factor analysis (Semi -SOFA) model is presented, further, to improve robustness, Semi-SOFA is extended to weighted form (WSemi-SOFA). This paper performs orthogonal decomposition on the obtained factors, which divides them into two parts: quality-related one and quality-unrelated one. Based on it, the corresponding $T^{2}$ statistics are designed to offer quality-related process monitoring, respectively. Besides, SPE statistics are constructed as supplement to monitor residuals. For effectiveness demonstration of the proposed method, TE benchmark is utilized.

[1]  Zhiqiang Ge,et al.  Robust supervised probabilistic principal component analysis model for soft sensing of key process variables , 2015 .

[2]  Jian Yang,et al.  Dynamic learning on the manifold with constrained time information and its application for dynamic process monitoring , 2017 .

[3]  Qinghua Li,et al.  Process monitoring based on factor analysis: Probabilistic analysis of monitoring statistics in presence of both complete and incomplete measurements , 2015 .

[4]  Okyay Kaynak,et al.  Improved PLS Focused on Key-Performance-Indicator-Related Fault Diagnosis , 2015, IEEE Transactions on Industrial Electronics.

[5]  Zhiqiang Ge,et al.  Probabilistic latent variable regression model for process-quality monitoring , 2014 .

[6]  Zhiqiang Ge,et al.  Supervised Latent Factor Analysis for Process Data Regression Modeling and Soft Sensor Application , 2016, IEEE Transactions on Control Systems Technology.

[7]  David J. Sandoz,et al.  The application of principal component analysis and kernel density estimation to enhance process monitoring , 2000 .

[8]  Tao Chen,et al.  Robust probabilistic PCA with missing data and contribution analysis for outlier detection , 2009, Comput. Stat. Data Anal..

[9]  Junghui Chen,et al.  Similarity based robust probability latent variable regression model and its kernel extension for process monitoring , 2017 .

[10]  ChangKyoo Yoo,et al.  Calibration, Prediction and Process Monitoring Model Based on Factor Analysis for Incomplete Process Data , 2005 .

[11]  Zhiqiang Ge,et al.  Semi-supervised PLVR models for process monitoring with unequal sample sizes of process variables and quality variables , 2015 .