Hybrid Model for Rolled Material Mechanical Property Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining mechanical properties of rolled materials, such as steel strips, face limitations due to the inability to accurately account for all process parameters, particularly cooling rate and location-specific variations, leading to unreliable predictions outside the trained data range and issues with missing or invalid measurements.
Innovation Solution
A hybrid model combining a physical production model and a trained statistical data model, where the physical model recreates the production process using metallurgical equations and the statistical model predicts mechanical properties based on learned correlations, with a weighting mechanism to account for the similarity of the input data to the training dataset.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If statistical data models are used to predict mechanical properties, then sampling complexity is reduced, but prediction reliability deteriorates when input data falls outside the trained data range
Solution Approach 1:
The patent combines a physical production model with a statistical data model into a hybrid system. The physical model ensures predictions remain reliable even when input data falls outside the trained statistical model's range, while the statistical model provides accurate predictions within the trained range. This merging resolves the contradiction by maintaining both operational ease and prediction reliability across all input conditions.
Solution Approach 2:
The patent introduces a weighting mechanism that dynamically changes parameters based on input data characteristics. When input data is within the trained range, the statistical model receives higher weight; when outside the range, the physical model receives higher weight. This parameter adaptation resolves the contradiction by optimizing prediction reliability for different input conditions while maintaining operational simplicity.
2Loss of information
If physical models are used to simulate production processes, then understanding of metallurgical properties is improved, but computational complexity increases
Solution Approach 1:
The patent applies partial action by using the physical model only when necessary (when statistical model predictions are unreliable or when deep metallurgical understanding is required). For routine predictions within the trained data range, the simpler statistical model is used. This resolves the contradiction by reducing computational complexity while maintaining metallurgical property understanding when needed.
Solution Approach 2:
The hybrid model structure acts as an intermediary between the physical and statistical models, selecting and weighting their outputs based on input conditions. This intermediary approach resolves the contradiction by allowing the system to benefit from metallurgical property understanding through the physical model only when necessary, while relying on the computationally efficient statistical model for routine operations.
3Measurement precision
If cooling rate and location-specific variations are fully accounted for, then mechanical property accuracy is improved, but measurement and data collection difficulty increases
Solution Approach 1:
The patent replaces direct physical measurement of cooling rates and location-specific variations with a computational approach. The hybrid model uses the physical production model to calculate these parameters based on available process data, and the statistical model to learn correlations from historical data. This substitution resolves the contradiction by achieving high measurement precision without the difficulty of direct physical measurement.
Data Source
AI summary
A method for determining mechanical properties of a first rolled material by a hybrid model that includes production datasets relating to further rolled materials, a physical production model and a statistical data model. The production dataset relating to the first rolled material is used to determine a first mechanical dataset, a further production dataset and a metallurgical dataset and also a second mechanical dataset. An averaged normalized distance value for production datasets relating to the further rolled materials is determined that is used to ascertain the mechanical properties of the rolled material as a weighted average from the first and second mechanical datasets. When creating the hybrid model, the physical production model is used to determine further production datasets relating to the further rolled goods for training the statistical data model.