Chemical Mixture Property Prediction Using Substance Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting properties of chemical mixtures, such as automotive paints, are labor-intensive due to complex relationships between mixture components and properties, lacking effective predictive models to estimate properties of new mixtures.
Innovation Solution
A computer-implemented method that trains a data-driven model by clustering similar ingredients into substance clusters, reducing complexity and using a rule-based machine learning model to predict properties, allowing for the estimation of new mixture properties based on these clusters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If laboratory personnel develop formulas through verification experiments to measure coating formulation properties, then the accuracy of property measurements is improved, but the time consumption and labor intensity increase significantly
Solution Approach 1:
The patent applies preliminary action by training predictive models in advance using historical formulation data and measured properties. Once trained, these models can rapidly predict properties of new formulations without requiring actual laboratory verification experiments, thus saving time while maintaining measurement accuracy through the model's predictions
Solution Approach 2:
The patent creates a virtual copy of the laboratory measurement process through computational models. Instead of physically measuring every new formulation in the lab, the system uses trained models that replicate the measurement function computationally, reducing the need for repeated physical experiments while maintaining prediction accuracy
2Reliability
If comprehensive verification experiments are conducted to measure all properties of chemical mixtures, then the reliability of property data is improved, but the complexity of the development process increases
Solution Approach 1:
The patent segments the complex development process into two distinct phases: (1) an initial training phase where models learn from historical data with measured properties, and (2) a prediction phase where trained models estimate properties of new formulations. This segmentation reduces overall process complexity by separating the computationally intensive model training from the rapid prediction process
Solution Approach 2:
The patent introduces trained predictive models as intermediaries between formulation composition and property measurement. These models act as mediators that translate ingredient information into property predictions, reducing the need for direct laboratory measurement while maintaining data reliability through the model's learned relationships
3Measurement precision
If detailed ingredient information is used in predictive models, then the precision of property predictions is improved, but the complexity of training datasets increases
Solution Approach 1:
The patent merges similar ingredients into ingredient clusters based on their chemical properties and functional characteristics. Instead of treating each individual ingredient separately, the system groups them into clusters that capture essential chemical similarities, reducing training dataset complexity while maintaining prediction precision through the clustered representation
Data Source
AI summary
Disclosed herein is a computer-implemented method for training a data-driven model for predicting properties of a chemical mixture. The method includes the steps of obtaining data including history and/or calibration data of a plurality of chemical mixture recipes and properties of each chemical mixture recipe, with each chemical mixture recipe including two or more ingredients, assigning at least one ingredient in each chemical mixture recipe to one of pre-defined substance clusters, each pre-defined substance cluster representing one ingredient or a group of ingredients having similar chemistry, revising each chemical mixture recipe by replacing the at least one ingredient with the assigned pre-defined substance cluster, and providing the revised chemical mixture recipes, together with the properties of the chemical mixture recipes, to a machine learning process in order to train a data-driven model, which is usable for predicting characteristics of properties of a new chemical mixture.