Material Condition Prediction Using Inverse Topological Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to predict the conditions for obtaining specific property values in materials development, particularly in materials like dental ceramics, using machine learning techniques.
Innovation Solution
A method involving topological data analysis, specifically persistent homology, is used to extract features from material images, followed by dimensionality reduction and machine learning to predict property values, and then perform inverse analysis to identify conditions for achieving those values, presenting results on different materials.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning is used to predict property values of materials, then prediction capability is improved, but the ability to predict conditions for obtaining specific property values deteriorates
Solution Approach 1:
The patent applies inverse analysis by inverting the forward prediction model. Instead of predicting property values from material conditions (forward analysis), the system predicts material conditions from desired property values (inverse analysis). This is achieved by training a model where the input-output relationship is reversed, allowing the system to work backwards from target properties to identify suitable material conditions, thus resolving the contradiction between prediction accuracy and condition prediction capability
2Measurement precision
If topological data analysis is applied to extract features from material images, then feature extraction accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent replaces traditional mechanical or manual feature extraction methods with topological data analysis, specifically persistent homology. This mathematical approach automatically extracts topological features from material images by analyzing the evolution of topological structures across different scales. The method substitutes complex manual processing with a systematic mathematical framework that provides both high accuracy and automation, reducing operational complexity while maintaining feature extraction precision
3Adaptability or versatility
If inverse analysis is performed on the prediction model, then condition prediction capability is improved, but computational requirements increase
Solution Approach 1:
The patent performs preliminary action by pre-training the inverse analysis model using computationally intensive procedures during the model development phase. The inverse model is trained offline using synthetic data generated from forward simulations, allowing the computationally expensive inverse analysis to be performed once during training. Once trained, the model can perform rapid condition predictions during actual use with minimal computational resources, thus resolving the contradiction between enhanced condition prediction capability and reduced operational computational requirements
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A condition for obtaining a property value of a target material is predicted. A method according to an embodiment of the present invention includes: acquiring a property value of a material; predicting a feature of the material based on the acquired property value of the material by performing an inverse analysis on a model that predicts a property value of the material based on a feature of the material; generating a result of topological data analysis based on the predicted feature of the material; and presenting, on an image of a material different from the material, a pixel corresponding to a portion designated in the result of the topological data analysis.