Polymer Composite Recipe Prediction With Feedback for Target Properties
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting polymer composite materials with target properties require extensive trial and error, are costly, time-consuming, and struggle to account for numerous variables and interactions, particularly when using recycled plastics.
Innovation Solution
A method and device utilizing a recipe prediction model trained on deep learning algorithms to predict and modify recipes for polymer composite materials, incorporating recycled polymers, by calculating and adjusting mixing ratios to achieve target properties, and providing an effective feedback loop for synthesis environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If trial and error methods are used to predict polymer composite materials with target properties, then material properties can be achieved, but development time and costs increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-training an AI model with extensive material composition data and property relationships before actual prediction. This pre-prepared knowledge base enables rapid prediction without requiring trial-and-error experiments during the actual material development process, thus reducing development time while maintaining prediction accuracy
Solution Approach 2:
The patent uses copying by creating a virtual model that replicates the complex relationships between material compositions and properties. Instead of physically testing multiple formulations, the system copies the essential patterns from training data into an AI model that can predict properties of new formulations, eliminating the need for repeated physical trials
2Extent of automation
If commercially available AI models are used for prediction, then some automation is achieved, but training costs and complexity increase
Solution Approach 1:
The patent applies parameter changes by adjusting the complexity and scope of the AI model based on specific application needs. Rather than using a fixed complex model, the system can modify parameters such as the types of properties predicted, the range of materials covered, and the depth of analysis to match the specific requirements, thereby reducing unnecessary complexity while maintaining automation
Solution Approach 2:
The patent uses segmentation by dividing the prediction task into modular components that can be independently trained and combined. The system can handle different material types, properties, and processing conditions as separate modules, allowing selective implementation based on needs and reducing overall system complexity while maintaining high automation capability
3Measurement precision
If numerous variables and interactions in recipes are considered, then prediction accuracy improves, but model complexity and training difficulty increase
Solution Approach 1:
The patent applies taking out by extracting and focusing on the most critical variables and interactions that have the greatest impact on material properties. Rather than attempting to model every possible variable, the system identifies and prioritizes the key factors, removing less significant elements to reduce model complexity while maintaining prediction precision for the most important properties
Data Source
Figure 1
Figure 2
Figure 3
AI summary
According to various embodiments, a method for generating recipes of a polymer composite material, which includes: acquiring a prediction recipe based on a preset target property of a target polymer composite material and a recipe prediction model; acquiring a result property of a polymer composite material generated based on the prediction recipe; calculating a difference value between the target property and the result property; and outputting or modifying the prediction recipe based on a result of comparing the difference value with a preset reference value, and a device thereof may be provided.