AI Chemical Formulation Prediction With Virtual Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional chemical formulation methods rely on time-consuming trial-and-error experiments and empirical rules, making it difficult to efficiently derive and validate chemical formulations with desired properties.
Innovation Solution
An AI-based automated chemical formulation apparatus and method using machine learning models to predict and validate chemical formulations, minimizing trial-and-error by employing a predicting model and a validation model to derive and verify formulations, followed by synthesis using a dispensing device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional trial-and-error experimental methods are used to derive chemical formulations, then researchers can obtain empirical data through actual experiments, but the process is time-consuming and requires extensive manual intervention
Solution Approach 1:
The system performs preliminary computational actions by training machine learning models on historical chemical formulation data before actual synthesis. The predicting model generates candidate formulations and the validation model pre-evaluates them computationally, allowing researchers to skip time-consuming trial-and-error experiments and directly synthesize only the most promising candidates.
Solution Approach 2:
The system creates virtual copies of chemical formulations through computational modeling. Instead of physically testing every possible formulation combination, the validation model creates and evaluates virtual representations of formulations, selecting only those with high predicted success rates for actual physical validation, thereby dramatically reducing experimental time while maintaining reliability.
2Productivity
If machine learning models are used to predict chemical formulations, then the formulation search becomes more efficient, but the models require training data and computational resources
Solution Approach 1:
The system segments the formulation discovery process into distinct functional modules: a predicting model generation unit that creates candidate formulations, a validation model generation unit that evaluates them, and a chemical dispensing device that executes synthesis. This segmentation allows each component to be optimized independently and simplifies the overall system architecture by assigning specific tasks to specialized units.
Solution Approach 2:
The validation model acts as an intermediary between the predicting model and the chemical dispensing device. It receives candidate formulations from the predicting model, computationally evaluates their likelihood of success, and filters them before passing selected candidates to the dispensing device for actual synthesis, thereby reducing the complexity of direct experimentation.
3Measurement precision
If physical property validation experiments are conducted for each formulated compound, then accurate property verification is achieved, but the process becomes expensive and time-consuming
Solution Approach 1:
Instead of conducting physical validation experiments on all possible formulations, the system applies partial action by performing computational validation on a large number of candidates and only executing physical experiments on a small subset of high-probability formulations selected by the validation model. This approach maintains measurement precision for the synthesized compounds while dramatically reducing overall experimental costs and energy consumption.
Data Source
AI summary
An automated chemical formulation apparatus includes: a data reception unit that receives an input chemical material dataset including chemical material information, chemical composition information, chemical formulation information and property information thereof; a predicting model generation unit that trains a first machine learning model using the input chemical material dataset to generate a predicting model for predicting a chemical formulation based on target property information of a target material; and a formulation prediction unit that sets a boundary condition based on the input chemical material dataset, generates a new input dataset including at least one of chemical material information, chemical composition information, and chemical formulation information within the boundary condition, inputs the new input dataset to the predicting model, and sets predetermined one or more pieces of target property information to perform prediction, thereby outputting a first group of chemical formulation data.