Machine Learning Material Characterization System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining material properties for structures, such as aircraft components, require extensive and time-consuming coupon testing, involving hundreds of thousands of tests, which is costly and inefficient.
Innovation Solution
A computer-based system that uses machine learning models trained with data from physical or virtual testing of coupons to estimate material properties, reducing the need for extensive testing by predicting properties for structures using augmented coupon test data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If coupon testing is performed to determine material properties, then accuracy of material property data is improved, but time consumption and cost increase significantly
Solution Approach 1:
The system performs preliminary actions by pre-testing a diverse set of coupons across multiple material types and conditions before the actual design process. This pre-characterization creates a training database that can be quickly queried during material selection, eliminating the need to perform extensive testing at the time of design while maintaining high accuracy through the comprehensive nature of the preliminary dataset.
Solution Approach 2:
The system creates virtual copies of physical coupon test results by using machine learning models to generate synthetic material property data based on limited physical testing. These virtual copies allow extensive material property exploration without repeating physical tests, thereby reducing time consumption while preserving measurement precision through the accuracy of the trained models.
2Reliability
If extensive coupon testing is performed to cover all material conditions, then reliability of material property data is improved, but productivity decreases due to the large number of tests required
Solution Approach 1:
The system performs preliminary comprehensive testing across diverse material types, processing methods, and environmental conditions to build a robust training database. This preliminary action ensures that the machine learning models are trained on reliable data covering the full range of possible material variations, thereby maintaining high reliability without requiring extensive testing during each design project.
Solution Approach 2:
The system changes parameters by using machine learning models to interpolate and extrapolate material properties across untested parameter combinations. Instead of physically testing every possible combination of material type, processing method, and environmental condition, the system learns the relationships between these parameters from limited physical tests and predicts properties for untested combinations, maintaining reliability while dramatically reducing the number of required tests.
3Productivity
If machine learning models are used to estimate material properties, then productivity is improved by reducing the number of tests, but device complexity increases due to the need for training data collection and model development
Solution Approach 1:
The system achieves universality by creating a single machine learning model framework that can estimate material properties across multiple material types (metals, polymers, ceramics, composites) and various processing methods. This multi-functional model reduces the need for separate testing and modeling systems for each material type, thereby improving productivity while managing complexity through a unified approach rather than multiple specialized systems.
Data Source
AI summary
A method, apparatus, system, and computer program product for estimating material properties. Training data comprising results of testing samples for a set of materials over a range of loads applied to the samples is identified by a computer system. A machine learning model is trained by the computer system to output the material properties for materials in structures using the training data.