Abnormality Cause Estimation in Reinforcement Learning Control

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In control systems utilizing multiple machine learning models, it is challenging to determine the main cause of abnormalities, as existing methods lack clarity in distinguishing between evaluation and operation models, leading to difficulties in identifying the root cause of issues.

Innovation Solution

An estimation apparatus and method that acquires an abnormality index from state data when an abnormality occurs, estimating whether the evaluation model or the operation model is the main cause based on this index, and outputs instructions to retrain either model accordingly.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple machine learning models (evaluation model and operation model) are used in a control system, then the system's functionality and performance are improved, but it becomes difficult to determine the main cause of abnormalities

Engineering Contradiction:
Improvesystem functionalityVSAvoidabnormality cause identification
Core Design Contradiction:
Adaptability or versatilityVSDifficulty of detecting and measuring

Solution Approach 1:

The patent segments the abnormality analysis by creating distinct abnormality indices for the evaluation model and operation model. The estimation unit divides the diagnostic task into separate evaluation paths, allowing independent analysis of each model's contribution to the abnormality through dedicated indices and threshold comparisons.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces abnormality indices as intermediary metrics that mediate between the complex multi-model system and the cause identification task. These indices serve as intermediate representations that translate model outputs into interpretable abnormality measurements, facilitating clearer cause determination.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If the system provides detailed abnormality analysis, then the accuracy of cause determination is improved, but the system complexity increases

Engineering Contradiction:
Improvecause determination accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the diagnostic system into distinct functional components: abnormality index acquisition units for each model type, an estimation unit for synthesis, and an output unit for results. This modular segmentation achieves precise cause determination while managing complexity through organized functional separation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent uses parameter changes in the form of abnormality indices and threshold values to achieve precise cause determination. By adjusting and comparing index values against predetermined thresholds, the system accurately identifies abnormality causes without requiring complex analytical structures.

Inventive Principle:
Principle #35Parameter changes

3Ease of operation

If the system automatically identifies the main cause of abnormalities, then the ease of operation is improved, but the reliability of the identification may be compromised

Engineering Contradiction:
Improveautomatic cause identificationVSAvoididentification reliability
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent implements feedback mechanisms where abnormality indices are continuously acquired and compared against thresholds, with results fed back to determine the main cause. This feedback loop enables automatic identification while maintaining reliability through systematic validation against predetermined criteria for each model type.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4296800A1Estimation apparatus, estimation method, and estimation program
Publication Date: 2023.12.27 YOKOGAWA ELECTRIC CORP
  • EP4296800A1 patent drawingFigure 1
  • EP4296800A1 patent drawingFigure 2
  • EP4296800A1 patent drawingFigure 3

AI summary

[Abstract][Means for solving the problem] There is provided an estimation apparatus including an abnormality index acquisition unit configured to acquire, as an abnormality index, an evaluation index which is output by an evaluation model according to state data being input when an abnormality occurs in a facility, among pieces of the state data indicating a state of the facility when an operation model is used to control a control target provided in the facility, the operation model being generated by reinforcement learning in which an output of the evaluation model trained by machine learning to output the evaluation index in accordance with the state of the facility, is set as at least a part of a reward, and outputting an action in accordance with the state of the facility; an estimation unit configured to estimate which of the evaluation model or the operation model is a main cause of the abnormality, based on the abnormality index; and an output unit configured to execute an output in accordance with a result of the estimate.