Abnormality Diagnosis Models for Multi-Facility Metal Processing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In manufacturing processes like steel production, existing methods struggle to accurately diagnose abnormalities from a large amount of sensor data, as many measured values are for operational control and do not directly indicate facility status or the cause of abnormalities, limiting the ability to cover all potential issues.

Innovation Solution

A construction method for an abnormality diagnosis model that creates two types of models: one learning relationships between measured values at the same time and another at the same position, using data from multiple facilities to diagnose abnormalities in rolling mills, allowing for early investigation of causes and efficient countermeasures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If a large number of sensors are used to measure operational data in the manufacturing process, then the quantity of measured values increases, but the ability to accurately diagnose abnormalities deteriorates because many measured values are for operational control and do not directly indicate facility status or cause of abnormality

Engineering Contradiction:

Improvequantity of measured valuesVSAvoidaccuracy of abnormality diagnosis

Core Design Contradiction:

Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent segments the large volume of measured values into two categories: (1) measured values for operational control and device control, and (2) measured values that indicate facility status or cause of abnormality. By separating these categories, the system can focus on the relevant measured values for abnormality diagnosis, thereby improving diagnostic accuracy despite having a large total quantity of measured values from multiple sensors.

Inventive Principle:

Principle #1Segmentation

2Reliability

If only model base approach is used to diagnose manufacturing status, then the theoretical framework is established, but the ability to cover all abnormality cases deteriorates due to the complexity and variation in manufacturing processes

Engineering Contradiction:

Improvetheoretical foundation of diagnosisVSAvoidcoverage of abnormality cases

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent merges the model base approach with the data base approach to create a hybrid diagnosis system. The model base approach provides the theoretical framework and physical/chemical phenomenon models, while the data base approach uses operational data to identify patterns and abnormalities. By combining both approaches, the system maintains theoretical reliability while gaining the adaptability to cover diverse abnormality cases that may not be predictable by models alone.

Inventive Principle:

Principle #5Merging (Combining)

3Loss of information

If data base approach using past trouble cases is used, then the ability to learn from historical data is improved, but the ability to predict unprecedented abnormalities deteriorates because aging facilities often experience novel troubles

Engineering Contradiction:

Improveutilization of historical dataVSAvoidprediction capability for new abnormalities

Core Design Contradiction:

Loss of informationVSReliability

Solution Approach 1:

The patent inverts the traditional data base approach by not only using past trouble cases but also incorporating normal operational data to establish baseline patterns. By comparing current operational data against both historical trouble cases and normal operation patterns, the system can detect deviations that indicate new or unprecedented abnormalities, thereby improving prediction capability while still utilizing historical data effectively.

Inventive Principle:

Principle #13The other way round (Inversion)

4Manufacturing precision

If measured values for operational control are prioritized, then the control accuracy is improved, but the ability to detect facility status and abnormality causes deteriorates because these measured values do not directly indicate such information

Engineering Contradiction:

Improveoperational control accuracyVSAvoiddetection of facility status

Core Design Contradiction:

Manufacturing precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The patent introduces an intermediary analysis layer that connects operational control measured values with facility status indicators. This intermediary layer processes and correlates the operational control data with other measured values to infer facility status and potential abnormality causes. This allows the system to maintain accurate operational control while simultaneously detecting facility status through the intermediary relationships between different measured values.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20230393113A1Construction method of abnormality diagnosis model, abnormality diagnosis method, construction device of abnormality diagnosis model, and abnormality diagnosis device

Publication Date: 2023.12.07 JFE STEEL CORP

AI summary

A construction method of an abnormality diagnosis model of a process for sequentially treating a metal material in a plurality of facilities, the construction method includes: creating a first abnormality diagnosis model that learns a relationship between measured values at a same time and an abnormality by using the measured values measured at the same time in a predetermined measurement cycle determined in advance for the plurality of facilities; and creating a second abnormality diagnosis model that learns a relationship between measured values at a same position and an abnormality by using the measured values at the same position of the metal material obtained by compiling the measured values measured in the plurality of facilities for each position of the metal material.