Fracture Surface Image Analysis for Rapid Polymer DBTT Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting the ductile-to-brittle transition temperature (DBTT) of polymer compositions are inefficient and resource-intensive, lacking a reliable computer-implemented approach that can accurately predict this critical material property without extensive experimentation.
Innovation Solution
A computer-implemented method using a machine learning algorithm trained with fracture surface images to predict DBTT, utilizing a convolutional neural network (CNN) to map data sets from training polymer compositions to DBTT values, allowing rapid prediction based on a single experiment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional experimental methods are used to determine DBTT, then measurement precision is improved, but loss of time and productivity deteriorate
Solution Approach 1:
The machine learning algorithm is trained in advance with fracture surface images and corresponding DBTT values from multiple polymer compositions. This preliminary training creates a predictive model that can rapidly estimate DBTT for new compositions without requiring time-consuming experimental measurements, thus resolving the contradiction between measurement precision and time consumption.
Solution Approach 2:
Instead of performing extensive experimental measurements on each new polymer composition, the system uses fracture surface images as copies or proxies for the actual material properties. The visual characteristics of fracture surfaces contain sufficient information to predict DBTT, replacing the need for lengthy experimental procedures while maintaining acceptable accuracy.
2Reliability
If extensive experimentation is conducted to determine DBTT, then reliability of DBTT determination is improved, but productivity deteriorates
Solution Approach 1:
The patent replaces the mechanical experimental system (physical testing of polymer compositions at various temperatures to determine DBTT) with an information-processing system based on machine learning. The algorithm processes fracture surface images and predicts DBTT values, substituting extensive physical experimentation with computational analysis that maintains reliability while dramatically improving productivity.
Solution Approach 2:
The approach changes the fundamental parameter used for DBTT determination from direct mechanical testing results to visual characteristics of fracture surfaces. By transforming the measurement parameter from a time-intensive experimental output to an image-based input, the system achieves reliable predictions without sacrificing productivity.
3Productivity
If a machine learning approach is used to predict DBTT, then productivity is improved, but measurement precision may deteriorate
Solution Approach 1:
The machine learning algorithm undergoes comprehensive preliminary training with a diverse dataset of fracture surface images and corresponding experimentally determined DBTT values. This extensive pre-training ensures that the model learns accurate relationships between visual features and DBTT, maintaining measurement precision while enabling rapid predictions for new compositions.
Solution Approach 2:
The system incorporates feedback mechanisms where the machine learning model's predictions can be validated and refined using experimental data. This feedback loop ensures that measurement precision is maintained or improved over time as the model learns from actual measurement outcomes, while the overall productivity benefit of rapid prediction is preserved.
Data Source
AI summary
The present invention relates to a computer implemented method for predicting the ductile-to-brittle transition temperature (DBTT) of a test polymer composition based on images. Furthermore, a non-transitory computer readable storage medium is provided for tangibly storing computer program instructions capable of being executed by a processor, the computer program instructions defining the steps of the aforementioned computer implemented method. Furthermore, the invention is directed to the use of fracture surface images and values indicative of the ductile-to-brittle transition temperatures (DBTT) of polymer compositions for training a machine learning algorithm.