Optimal Transport Image Representation and Deep Covariance Alignment (CORAL) for Control Valve Stiction Detection

2026-07-24 · Source: Computer Vision and Pattern Recognition · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Expert, quick

Summary

A novel methodology for control valve stiction detection combines optimal transport (OT) imaging with deep correlation alignment (Deep CORAL) to address domain shift between simulated and real industrial data. This approach converts closed-loop signals, specifically controller output and process variable, into two-dimensional OT images that capture dynamic control loop behavior. A convolutional neural network (CNN) encoder is trained using a dual objective: cross-entropy loss on labeled simulation data and a Deep CORAL loss to align covariance between simulation and unlabeled industrial data, thereby learning domain-invariant representations. When evaluated on an independent test set of 20 industrial stiction benchmark loops, the methodology successfully diagnosed 18 loops, achieving 100% recall across all 13 stiction cases, 90.00% accuracy, and a 92.86% F1-score. This method significantly outperforms standard baseline approaches by providing robust and reliable stiction detection.

Key takeaway

For Machine Learning Engineers developing diagnostic models for industrial control systems, this methodology offers a robust solution to the persistent challenge of domain shift. You should consider integrating optimal transport (OT) imaging for dynamic signal representation and Deep CORAL for unsupervised domain adaptation. This approach, which achieved 90.00% accuracy and 100% recall in stiction detection, can significantly improve the reliability and generalization of your models across diverse real-world industrial data.

Key insights

Combining optimal transport imaging with Deep CORAL effectively mitigates domain shift for robust control valve stiction detection in industrial settings.

Principles

• Domain shift is a key challenge for data-driven industrial diagnostics.
• Optimal Transport images capture dynamic system behavior effectively.
• Covariance alignment can create domain-invariant feature representations.

Method

Convert control loop signals into 2D Optimal Transport images. Train a CNN encoder with cross-entropy loss on labeled simulation data and Deep CORAL loss for covariance alignment between simulation and unlabeled industrial data to learn domain-invariant features.

In practice

• Apply OT imaging to time-series data for dynamic feature extraction.
• Use Deep CORAL for unsupervised domain adaptation in industrial AI.
• Implement combined loss functions for robust model generalization.

Topics

• Control Valve Stiction
• Optimal Transport
• Deep CORAL
• Domain Adaptation
• Convolutional Neural Networks
• Industrial Diagnostics

Best for: Research Scientist, AI Scientist, Machine Learning Engineer

Related on AIssential

• Optimal Transport Image Representation and Deep Covariance Alignment (CORAL) for Control Valve Stiction Detection — Optimal Transport imaging combined with Deep CORAL effectively mitigates domain shift for robust control valve stiction detection in industrial settings.
• Why Domain Matters: A Preliminary Study of Domain Effects in Underwater Object Detection — A new framework defines underwater domains using measurable image, scene, and acquisition properties for robust object detection evaluation.
• Geometric Dictionary Learning of Dynamical Systems with Optimal Transport — DOODL learns shared spectral manifolds of dynamical systems for robust, data-efficient operator estimation and system identification.
• Latent Variable-Mediated Cross-Learning for Few-Shot Acoustic Impedance Imaging — RD-SCL integrates physics-guided deconvolution with semi-supervised cross-learning for robust few-shot acoustic impedance imaging.
• Semantically Aware Optimal Transport for Dense Label Transfer — SAOT improves dense label transfer across visual domains by learning semantic transport costs with unbalanced optimal transport.
• Neptuna: A Comprehensive Machine Learning Framework for Benchmarking Complex Multiphase Flows — Neptuna benchmarks shock-driven multiphase flows, showing composite losses and SoftAdapt improve ML surrogate fidelity, despite no single model excelling.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computer Vision and Pattern Recognition.