Adaptive Model Learning With Ground Truth Profile Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data-driven and physics-based models in manufacturing systems face accuracy issues due to corrupted ground truth data, particularly in industrial sintering plants, where wear and tear lead to noisy wind box temperature profiles, affecting the models' ability to adapt to new operating regimes and learn underlying physical processes effectively.
Innovation Solution
A processor-implemented method and system for adaptive learning that corrects distorted ground truth profiles using a Profile Deviation Index (PDI) and iteratively refines them to meet quality thresholds, integrating simulated data from soft sensors to improve data quality, and selects relevant instances for model retuning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If ground truth data is used for model retuning, then model adaptability to new operating regimes is improved, but model accuracy deteriorates due to corrupted ground truth data
Solution Approach 1:
The patent introduces an intermediary system consisting of multiple soft sensors (physics-based and data-driven) that mediate between the corrupted ground truth measurements and the model retuning process. These soft sensors process and refine the ground truth data through ensemble methods, acting as a buffer that filters out corruption while preserving the essential information needed for model adaptation to new operating regimes.
Solution Approach 2:
The patent creates multiple copies of the ground truth measurement through different soft sensor models (physics-based soft sensors and data-driven soft sensors). By generating multiple representations of the same physical quantity through different modeling approaches, the system can compare and reconcile these copies to identify and correct corrupted data, thereby maintaining accuracy while enabling model adaptability.
2Adaptability or versatility
If ground truth data with air ingression noise is used for model learning, then model can adapt to current operating conditions, but learning effectiveness deteriorates due to noisy data
Solution Approach 1:
The patent implements feedback mechanisms where the ensemble of soft sensors continuously monitors and evaluates the quality of ground truth data. The system uses the outputs from multiple soft sensor models to provide feedback on data reliability, identifying when air ingression or other noise sources are corrupting measurements. This feedback loop enables the system to selectively use high-quality data for model learning, maintaining both adaptability and reliability.
3Measurement precision
If multiple soft sensors are used to correct ground truth data, then data accuracy is improved, but system complexity increases
Solution Approach 1:
The patent designs soft sensors that serve multiple functions: they can estimate physical quantities from limited measurements, detect data corruption, provide uncertainty quantification, and contribute to ensemble-based correction. By making each soft sensor multi-functional, the system achieves high data accuracy without proportionally increasing complexity, as the same computational resources serve multiple purposes in the data correction pipeline.
Data Source
AI summary
In applications such as adaptive learning of physics-based and data-driven models associated with industrial plants, the models are corrected periodically by taking into consideration the dynamic changes occurring in plant conditions and related data. However, accuracy of adaptive learning depends on accuracy of ground truth data being used as reference data. The disclosure herein generally relates to data preprocessing, and, more particularly, to a method and system for ground truth profile correction and instance selection. The system performs a ground truth profile correction for ground truth profiles having a Profile Deviation Index (PDI) value exceeding a threshold of distortion, to reduce the PDI value, and in turn reduce the distortion in the ground truth profiles. Further, the system performs a data instance selection to identify and remove outliers, and the data that remains after the data instance selection may be then used for applications such as model generation or retuning.