Elastic Strain Modeling for 3D Material Property Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current strain engineering of materials is limited to uniaxial and biaxial strains due to the complexity in predicting and testing the entire range of possible strains and resulting material properties, making it impractical to explore three-dimensional or higher-dimensional strain spaces effectively.
Innovation Solution
Development of trained statistical models for predicting bandgap and energy dispersion using three-dimensional and six-dimensional strain tensors, utilizing machine learning techniques to generate models from training data, allowing for the determination of material properties under various strain conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If trained statistical models using machine learning are developed to predict bandgap and energy dispersion across three-dimensional and six-dimensional strain spaces, then the ability to determine material properties under various strain conditions is significantly improved, but the complexity of the prediction system increases
Solution Approach 1:
The patent introduces trained statistical models as intermediary tools between strain input and material property prediction. These models act as mediators that have been pre-trained on comprehensive strain-space data, allowing complex multi-dimensional strain relationships to be captured without requiring real-time complex calculations. The models serve as computational intermediaries that simplify the prediction process while maintaining high accuracy across three-dimensional and six-dimensional strain spaces.
Solution Approach 2:
The patent applies preliminary action by pre-training statistical models on extensive training data that covers the entire range of possible strains before actual material property determination is needed. This pre-computation approach allows the models to learn complex strain-property relationships in advance, so that during actual use, predictions can be made quickly without performing complex real-time analyses. The training phase performs the heavy computational lifting beforehand.
2Measurement precision
If the entire range of possible strains is experimentally tested to predict material properties, then measurement precision is improved, but the time and resources required increase significantly
Solution Approach 1:
The patent creates computational copies of material behavior through trained statistical models that replicate the complex relationships between strain and material properties. Instead of physically testing every possible strain condition, the models serve as virtual copies that have been trained on representative data and can predict outcomes for any strain state within the trained range. This copying approach maintains measurement precision while eliminating the need for exhaustive physical experimentation.
Solution Approach 2:
The patent transforms the problem from physical experimentation to computational prediction by changing the parameters from physical strain applications to statistical model inputs. The statistical models accept strain coordinates as parameters and output predicted material properties, effectively changing the domain from experimental physics to computational statistics. This parameter transformation allows rapid exploration of strain space without physical constraints.
Data Source
AI summary
Methods for training statistical models for the bandgap and energy dispersion of materials as a function of an applied strain, as well as uses of these trained statistical models for elastic strain engineering of materials, are described.