Plant Abnormality Diagnosis Using Predicted Symptom Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing monitoring systems for plant abnormalities face high computation loads and inefficient diagnosis processes due to the lack of consideration for behavior patterns in abnormality signals, leading to increased diagnosis time and load.
Innovation Solution
An abnormality diagnosis system that predicts the development of instrument parameters using extrapolation to generate symptom motion patterns, which are then matched against stored abnormality model patterns to rapidly identify the cause of abnormalities, utilizing a Bayesian network for high accuracy and reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If failure diagnosis is performed on the basis of all abnormality signals without considering behavior patterns, then comprehensive diagnosis coverage is achieved, but diagnosis process load increases and efficiency decreases
Solution Approach 1:
The patent segments the diagnosis process by classifying abnormality signals into different behavior patterns (monotonic increase, monotonic decrease, periodic, random). This segmentation allows the system to handle different types of abnormalities separately using appropriate diagnosis methods for each pattern, rather than processing all signals uniformly, thereby reducing overall process load while maintaining comprehensive coverage.
Solution Approach 2:
The patent changes the parameter of signal processing by applying different analysis methods based on the behavior pattern parameter. For example, monotonic signals are analyzed using trend analysis, while periodic signals use spectral analysis. This parameter-based differentiation optimizes the diagnosis process by matching the analysis method to the signal characteristics, reducing unnecessary computational complexity.
2Measurement precision
If Mahalanobis distance calculation is used to detect abnormality predict, then abnormality detection accuracy is improved, but computation load increases
Solution Approach 1:
The patent applies partial action by using Mahalanobis distance calculation only for specific types of abnormality detection where it provides significant accuracy improvement, rather than applying it universally to all monitoring scenarios. For routine monitoring, simpler threshold-based methods are used, reserving the computationally intensive Mahalanobis distance for cases where higher precision is critical.
3Reliability
If all abnormality signals are used for failure diagnosis, then diagnosis completeness is maintained, but diagnosis time increases
Solution Approach 1:
The patent performs preliminary classification of abnormality signals into behavior patterns before the actual failure diagnosis process. By pre-categorizing signals based on their behavior characteristics (monotonic, periodic, random), the system prepares the data in advance for targeted diagnosis methods, reducing the time required during the actual diagnosis phase while ensuring all relevant signals are considered.
Data Source
AI summary
This abnormality diagnosis system for diagnosing abnormalities in a plant is provided with: an abnormality diagnosis control unit which, with respect to an instrument parameter measured in a plant determined to have an indication of abnormality, predicts the development of the instrument parameter by extrapolation, and which generates an abnormality manifestation pattern that is a pattern of behavior of the instrument parameter after prediction; and a data storage unit which stores a plurality of abnormality model patterns PA, PB that are patterns of behavior of the instrument parameters corresponding to causes CA1, CA2, CB1, CB2 of plant abnormality. The abnormality diagnosis control unit makes a matching determination between the abnormality manifestation pattern that has been generated and the plurality of abnormality model patterns PA, PB stored in the data storage unit, and identifies, as the cause of the abnormality in the abnormality manifestation pattern.