Adaptive Model Training for Power Plant Asset Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing model training for power plant equipment is static and fails to adapt to asset aging or changing operating conditions, leading to false alarms during on-line monitoring due to poor data estimation, which is a common issue in the power industry.
Innovation Solution
An adaptive model training system and method that selectively calibrates models using asset operating data, filtering out abnormal data to maintain accurate representations of normal operation, allowing for dynamic recalibration without manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a model is statically calibrated prior to on-line monitoring, then the model can be deployed for monitoring, but the model fails to adapt to asset aging or operating condition changes, leading to false alarms
Solution Approach 1:
The patent transforms the static model calibration approach into a dynamic one by implementing continuous or periodic recalibration of the model during on-line monitoring operations. The model is automatically retrained using incoming operational data, allowing it to adapt to asset aging and changing operating conditions while maintaining accurate condition assessment and reducing false alarms.
2Measurement precision
If machine learning is used to calibrate the model, then the model can learn normal operation patterns, but manual recalibration is required which reduces productivity
Solution Approach 1:
The patent implements an automated self-calibration system where the model continuously retrains itself using operational data without requiring manual intervention. The system automatically identifies when recalibration is needed, selects appropriate training data, and updates the model parameters, thereby maintaining high measurement precision while eliminating the productivity loss associated with manual recalibration tasks.
Solution Approach 2:
The system incorporates feedback mechanisms that monitor model performance during on-line operations and automatically trigger recalibration when performance degradation is detected. This closed-loop approach ensures the model maintains accuracy while adapting to changing conditions, and the automated feedback-driven process eliminates the need for manual monitoring and intervention.
3Adaptability or versatility
If the model is continuously updated with new data, then the model adapts to changing conditions, but abnormal data may be incorporated which degrades model performance
Solution Approach 1:
The patent applies preliminary filtering and validation actions to incoming operational data before it is used for model recalibration. The system pre-processes data to identify and exclude abnormal or erroneous measurements, ensuring that only high-quality representative data is incorporated into the training set. This preliminary data curation maintains model accuracy while enabling continuous adaptation to legitimate changing conditions.
Data Source
AI summary
An adaptive model training system and method for filtering asset operating data values acquired from a monitored asset for selectively choosing asset operating data values that meet at least one predefined criterion of good data quality while rejecting asset operating data values that fail to meet at least the one predefined criterion of good data quality; and recalibrating a previously trained or calibrated model having a learned scope of normal operation of the asset by utilizing the asset operating data values that meet at least the one predefined criterion of good data quality for adjusting the learned scope of normal operation of the asset for defining a recalibrated model having the adjusted learned scope of normal operation of the asset.