Related Experiment Video
Updated: Jul 12, 2025
08:12
A Psychophysics Paradigm for the Collection and Analysis of Similarity Judgments
Published on: March 1, 2022
2.5K
Visual Process Monitoring by Data-Dependent Kernel Discriminant Analysis with t-Distributed Similarities
Chihang Wei1,2, Chenglin Wen1, Jieguang He1
1Guangdong Key Laboratory of Petrochemical Equipment Fault Diagnosis, Guangdong University of Petrochemical Technology, Maoming 525000, China.
ACS Omega
|October 23, 2023
Summary
This study introduces Data-Dependent Kernel Discriminant Analysis (D²K-DA) for enhanced visual process monitoring. The method improves fault detection and process understanding through advanced data-driven visualization techniques.
Area of Science:
- Process Engineering
- Data Science
- Machine Learning
Background:
- Visual process monitoring offers intuitive insights into operational status and fault occurrences.
- Conventional methods have limitations in effectively visualizing complex process data.
Purpose of the Study:
- To propose a novel model for enhanced visual process monitoring.
- To improve the interpretability and fault detection capabilities of process monitoring systems.
Main Methods:
- Development of the Data-Dependent Kernel Discriminant Analysis (D²K-DA) model.
- Construction and learning of a data-dependent kernel function.
- Innovative optimization exploiting discriminative information and geometric similarities.
Main Results:
- Achieved low-dimensional visualizations with guaranteed intraclass compactness and interclass separability.
- Demonstrated preservation of both local and global geometry.
- Validated through experiments on simulated and real-life industrial processes.
Conclusions:
- The proposed D²K-DA model significantly enhances visual process monitoring.
- The method provides clear and comprehensible information for process status and fault analysis.
- Effective for both simulated and real-world industrial applications.
