Facility Control Loop Abnormality Detection Using CNN Time-Series Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Control valves in industrial facilities suffer from valve non-linearities such as stiction, leading to oscillations, product quality variation, and control-system instability, which are difficult to detect efficiently and non-invasively.

Innovation Solution

A convolutional neural network (CNN) based method processes time series data directly from control loops to detect valve stiction and other abnormalities, using labeled data and feedback from control engineers to improve accuracy and robustness, enabling rapid detection without disrupting plant processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional PV/SP/OP plot analysis is used to detect valve stiction, then detection accuracy can be achieved, but the detection process is time-consuming and requires manual expert analysis

Engineering Contradiction:
Improvestiction detection accuracyVSAvoiddetection time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces the manual mechanical analysis process with an automated electronic system. A convolutional neural network (CNN) model is trained to automatically analyze control loop data and detect valve stiction, eliminating the need for manual plot inspection by control engineers. The system processes time-series data from control loops and automatically identifies stiction patterns, achieving both high accuracy and rapid detection.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If invasive methods are used to detect valve abnormalities, then accurate diagnosis can be achieved, but the plant processes are disrupted

Engineering Contradiction:
Improveabnormality detection accuracyVSAvoidprocess continuity
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent introduces an intermediary automated analysis system that processes control loop data without physically interacting with the valve or process equipment. The CNN model analyzes existing process data (PV, SP, OP signals) to detect abnormalities, serving as a non-invasive mediator that provides accurate diagnosis while maintaining uninterrupted plant operations.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Measurement precision

If manual analysis by control engineers is performed, then detection accuracy is maintained, but productivity and efficiency are reduced

Engineering Contradiction:
Improvestiction detection accuracyVSAvoiddetection throughput
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent implements a self-service automated detection system where the CNN model independently analyzes control loop data and identifies valve stiction without requiring control engineer intervention. The system automatically processes multiple control loops, generates detection results, and can prioritize alerts, significantly increasing detection throughput while maintaining accuracy comparable to expert analysis.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4172705B1Machine learning based direct method of determining status of facility control loop components
Publication Date: 2026.04.01 HONEYWELL INTERNATIONAL INC
  • EP4172705B1 patent drawingFigure 1A
  • EP4172705B1 patent drawingFigure 1B
  • EP4172705B1 patent drawingFigure 2

AI summary

A trained machine learning algorithm processes time series production data. The time series production data are representative of a control process within a facility control loop. The machine learning training algorithm is trained using positive training data that are representative of a normal operation of components within the facility control loop and negative training data that are representative of an abnormal operation of components within the facility control loop. Output of the trained machine learning algorithm identifies abnormalities in the facility control loop.