Plant Sensor Monitoring for High-Low Abnormality Separation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The Mahalanobis Taguchi method struggles to distinguish between high value and low value abnormalities in plant sensors, as the Mahalanobis distance increases for both high and low sensor values, making it difficult to accurately identify the cause of abnormalities.
Innovation Solution
A plant monitoring device that acquires sensor bundles, calculates Mahalanobis distances, and determines whether the increased distance is due to high or low values, using a failure part estimation database to differentiate between likely and unlikely abnormality causes, thereby enhancing the reliability of abnormality cause estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the Mahalanobis Taguchi method is used to monitor plant operation state, then the abnormality detection capability is improved, but the ability to distinguish between high value and low value abnormalities deteriorates
Solution Approach 1:
The patent segments the abnormality analysis by creating separate evaluation processes for high value abnormalities and low value abnormalities. After detecting an abnormality using the Mahalanobis distance, the system divides the analysis into two distinct paths: one for evaluating sensors showing high values and another for sensors showing low values, thereby resolving the inability to distinguish between the two types of abnormalities.
Solution Approach 2:
The patent applies local quality by tailoring the evaluation method to the specific nature of each sensor reading. For sensors with high values, the system evaluates based on the assumption of high value abnormality, while for sensors with low values, it evaluates based on low value abnormality assumptions. This localized evaluation approach improves measurement precision for each specific case.
2Reliability
If the Mahalanobis distance is calculated for each sensor to identify abnormal sensors, then the abnormality detection is improved, but the accuracy of cause estimation deteriorates due to inability to distinguish high value/low value abnormalities
Solution Approach 1:
The patent segments the cause estimation process into distinct evaluation paths for high value and low value abnormalities. By calculating Mahalanobis distances separately for each type of abnormality and evaluating causes based on the specific type detected, the system preserves critical information that would otherwise be lost in a unified evaluation approach.
Solution Approach 2:
The patent adds a new dimension to the evaluation by considering both the magnitude of the Mahalanobis distance and the direction (high value or low value) of the sensor reading. This dimensional expansion allows the system to maintain accurate cause estimation by incorporating the qualitative aspect of the abnormality type into the quantitative distance measurement.
Data Source
AI summary
An acquisition unit acquires a bundle of detection values for each of a plurality of sensor values pertaining to a plant. A distance calculation unit obtains the Mahalanobis distance of the bundle of detection values acquired by the acquisition unit using, as reference, a unit space constituted by a collection of bundles of detection values for each of the plurality of sensor values. A determining unit determines, based on whether the Mahalanobis distance is at or within a prescribed threshold, whether the operation state of the plant is normal or abnormal. A trend specification unit specifies a trend with regards to at least one sensor value. An abnormality cause estimation unit estimates an abnormality cause based on the trend for the sensor value(s), and a fault site estimation database for holding the relationship between abnormality causes that may occur in the plant and sensor values for each of the trends.