Plant Abnormality Diagnosis Using ART Distance-Based Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing plant abnormality diagnosis systems using Adaptive Resonance Theory (ART) face delays in detecting abnormalities in entire plants due to the need for convergent calculation with each sensor measurement, leading to potential missed detection of sudden abnormalities.

Innovation Solution

A plant abnormality diagnosis device and system that calculates an abnormality rate for the entire plant from multiple sensor measurements using ART, determining abnormality based on distance differences in a multidimensional space, and controls the timing of ART processing to reduce calculation load and improve detection speed.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If ART processing is performed for each sensor measurement to determine abnormality, then measurement precision is improved, but loss of time increases due to convergent calculation requirements

Engineering Contradiction:
Improveabnormality detection accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent pre-calculates and stores reference values representing normal plant operation states before actual monitoring begins. During operation, the system compares current sensor measurements against these pre-established references using simple distance calculations in multidimensional space, avoiding the need for time-consuming convergent calculations at each measurement point while maintaining accurate abnormality detection

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent divides the plant monitoring into independent dimensional parameters, where each sensor measurement corresponds to a specific dimension. The system calculates distance differences along each dimension separately and aggregates them to determine overall abnormality, allowing parallel processing and reducing computational time compared to holistic convergent analysis

Inventive Principle:
Principle #1Segmentation

2Productivity

If ART processing is controlled based on abnormality rates and data accumulation, then productivity is improved through reduced calculation load, but measurement precision may deteriorate due to less frequent processing

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidabnormality detection accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements dynamic processing control where the frequency and depth of ART processing adapt based on the calculated abnormality rate. When abnormality rates are low and data accumulation is sufficient, processing is reduced to maintain high productivity. When abnormality rates increase or data accumulation is insufficient, processing frequency increases to maintain detection accuracy, creating a dynamic balance between efficiency and precision

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3514642B1Plant abnormality diagnosis device and plant abnormality diagnosis system
Publication Date: 2021.06.09 HITACHI LTD
  • EP3514642B1 patent drawingFigure 1
  • EP3514642B1 patent drawingFigure 2
  • EP3514642B1 patent drawingFigure 3~4

AI summary

Provided are a plant abnormality diagnosis device and a plant abnormality diagnosis system capable of determining an abnormality at an early stage in a plant as a whole during plant abnormality diagnosis using adaptive resonance theory (ART). A plant abnormality diagnosis device 2 is provided with an abnormality degree calculation unit 103 which, with respect to a plurality of pieces of measurement data from various sensors installed in the plant 5 to be diagnosed, determines an abnormality degree A of a plant 5 to be diagnosed, as a whole, on the basis of a difference in spatial distance between data belonging to a category determined by ART using data at a normal time and the plurality of pieces of measurement data.