Quality-Ensured In-Situ Process Monitoring with Deep Canonical Correlation Analysis

2025-09-23

AI Summary

Modeling nonlinear associations between high-dimensional process signals (e.g., optical emission spectroscopy, CT images) and offline quality measurements remains challenging; conventional control charts rely on Gaussian assumptions and require labeled defective samples. Method: This paper proposes an unsupervised, in-situ quality monitoring framework based on Deep Canonical Correlation Analysis (DCCA), which learns joint multimodal representations to automatically capture nonlinear covariation structures between process features and quality indicators—without assuming data normality or requiring defect labels. Theoretical guarantees on convergence and stability are provided. Results: Evaluated on direct metal deposition additive manufacturing, the method significantly improves monitoring accuracy and early anomaly detection capability compared to classical control chart approaches.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Computer Vision: Multi-modal Vision

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Web data quality in the era of algorithmically-generated content

Abstract

This paper proposes a deep learning-based approach for in-situ process monitoring that captures nonlinear relationships between in-control high-dimensional process signature signals and offline product quality data. Specifically, we introduce a Deep Canonical Correlation Analysis (DCCA)-based framework that enables the joint feature extraction and correlation analysis of multi-modal data sources, such as optical emission spectra and CT scan images, which are collected in advanced manufacturing processes. This unified framework facilitates online quality monitoring by learning quality-oriented representations without requiring labeled defective samples and avoids the non-normality issues that often degrade traditional control chart-based monitoring techniques. We provide theoretical guarantees for the method's stability and convergence and validate its effectiveness and practical applicability through simulation experiments and a real-world case study on Direct Metal Deposition (DMD) additive manufacturing.

Problem

• Capturing nonlinear relationships between process signals and product quality
• Enabling joint feature extraction from multi-modal manufacturing data sources
• Facilitating online quality monitoring without labeled defective samples

Innovation

• Deep Canonical Correlation Analysis for multi-modal data
• Joint feature extraction without labeled defective samples
• Online quality monitoring overcoming non-normality issues