Polymer Property Prediction Using Multi-Group Molecular Descriptors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods struggle to accurately predict the physical properties of polymers due to the complexity of molecular structures, which are not fully represented by individual descriptors, and lack sufficient correlations between calculated and measured properties, limiting the ability to determine suitable applications and generate comprehensive polymer databases.
Innovation Solution
A computer-based system that receives descriptors from multiple groups (atomic species count, functional count, monomer species count, etc.) to determine prediction scores for target polymers based on pre-existing polymer data, providing physical property data and formulas for raw materials, thereby enhancing prediction accuracy and database generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If individual molecular descriptors are used to represent polymer structure, then the representation is simple, but the accuracy of predicting physical properties deteriorates due to insufficient capture of molecular complexity
Solution Approach 1:
The patent segments the molecular structure representation into multiple descriptor groups (atomic species, functional groups, monomer species, bonds, morphology, topology, stereochemistry, etc.), where each group captures specific structural aspects. This segmentation allows comprehensive representation of polymer complexity while maintaining organized, manageable descriptor sets that can be systematically applied to predict physical properties.
Solution Approach 2:
The patent creates a composite descriptor system that integrates multiple types of molecular descriptors (atomic, functional, monomer, bond, morphology, topology, stereochemistry) into a unified representation framework. This composite approach combines the strengths of different descriptor types to achieve accurate physical property predictions while systematically representing the complex polymer structure.
2Measurement precision
If comprehensive descriptor groups are used to accurately represent polymer structure, then prediction accuracy improves, but the complexity of the system increases
Solution Approach 1:
The patent divides the comprehensive descriptor system into seven distinct descriptor groups (atomic species, functional groups, monomer species, bonds, morphology, topology, stereochemistry), each with specific sub-descriptors. This segmentation reduces system complexity by organizing numerous descriptors into structured categories, making the comprehensive representation more manageable and systematically applicable.
Solution Approach 2:
The patent introduces multiple dimensions of structural representation by adding descriptor groups across different structural levels (atomic → functional → monomer → bond → morphology → topology → stereochemistry). This dimensional expansion allows comprehensive property prediction without overwhelming complexity, as each dimension addresses specific structural aspects independently.
3Measurement precision
If manual measurements are used to determine polymer properties, then the data is accurate, but the productivity and ability to generate comprehensive databases deteriorates
Solution Approach 1:
The patent replaces manual measurement mechanisms with a computational prediction system that uses molecular descriptors and machine learning models to calculate physical properties. This substitution eliminates the need for time-consuming manual experiments while maintaining accurate property determination, thereby dramatically improving productivity and enabling comprehensive database generation.
Solution Approach 2:
The patent performs preliminary computational analysis by calculating molecular descriptors from polymer structure data before physical property prediction. This preliminary structuring of molecular information enables efficient batch processing of multiple polymers, allowing comprehensive database generation without repeated manual measurements for each polymer.
Data Source
AI summary
A system, method, and computer program product for predicting a polymer. A method may include determining one or more prediction scores for a target polymer based on at least one descriptor of the target polymer from at least three descriptor groups, wherein the one or more prediction scores include a prediction of one or more physical properties for the target polymer. A method may include determining one or more prediction scores for a target polymer based on at least one desired physical property of the target polymer, wherein the one or more prediction scores include a prediction of one or more raw materials of the target polymer.