Quality-related process monitoring scheme based on neighborhood embedding canonical correlation analysis model

Background: Accompanied by the development of sensor technology and the scale and integration of industrial processes, the safety and quality of the operating process is widely concerned. Process monitoring is an important technology for modern enterprises to ensure product quality and improve comprehensive economic efficiency. Methods: Canonical correlation analysis (CCA) is a useful algorithm for exploring the correlation between two sets of variables and has been successfully employed in quality-related process monitoring. Nevertheless, CCA neglects the neighborhood structure information while capturing the global maximum correlation feature. To ensure a more comprehensive feature representation, considering the effectiveness of neighborhood preserving embedding (NPE) algorithm in extracting local structure, this work proposes a quality-related process monitoring model named Neighborhood Embedding Canonical Correlation Analysis (NECCA). Firstly, the neighborhood information extracted through the improved NPE algorithm is incorporated into CCA. This model not only possesses analogous expression and analytical solution with CCA but also integrates the local structural feature. Secondly, a regression model is established, then the coefficient matrix is decomposed to distinguish qualityrelated and quality-unrelated subspaces. Finally, the proposed model is evaluated in a typical test case to demonstrate its rationality and effectiveness.

Automated summary

With the development of sensor technology and industrial processes, process monitoring has become crucial for ensuring product quality and improving economic efficiency. This study proposes a model called Neighborhood Embedding Canonical Correlation Analysis (NECCA) that combines canonical correlation analysis (CCA) with a neighborhood structure feature extraction algorithm. By incorporating neighborhood information into the traditional CCA model, the NECCA model achieves a more comprehensive feature representation. The rationality and effectiveness of the proposed model are demonstrated through a typical test case.