Automated Material Composition And Process Design for Target Properties
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automated materials design techniques fail to integrate material composition and manufacturing processing steps, neglect multiple manufacturing routes, and overlook manufacturability, ESG norms, and cost considerations, leading to inefficient and resource-intensive material development processes.
Innovation Solution
A processor-implemented method using reinforcement learning (RL) models to predict material compositions and manufacturing process parameters, integrating composition selection and sequential manufacturing steps, while considering ESG norms and cost, through a modular multi-agent setup that learns from historical data and computational models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional automated materials design techniques focus on correlating composition and properties or use single-process models, then the design process is simplified, but the integration of material composition and manufacturing processing steps is lost, leading to incomplete design solutions
Solution Approach 1:
The patent merges multiple separate design processes (composition design, process parameter design, structure prediction, property evaluation) into a single integrated automated workflow. The system combines machine learning models for composition-property relationships with process simulation models to simultaneously optimize material composition and manufacturing parameters, ensuring both simplification and completeness of the design solution.
Solution Approach 2:
The patent creates a universal automated design platform that handles multiple material systems and manufacturing processes through a single integrated system. The framework uses multi-functional models that can predict both material structure and properties while considering various manufacturing constraints, making the system applicable to diverse materials and processes without requiring separate design approaches for each case.
2Manufacturing precision
If extensive experimental trials are conducted for materials design, then accurate material properties are achieved, but the process becomes time-consuming and resource-intensive
Solution Approach 1:
The patent performs preliminary computational screening and prediction of material compositions and properties before conducting experimental trials. The system uses trained machine learning models to predict which compositions are most likely to achieve target properties, allowing researchers to prioritize only the most promising candidates for experimental validation, thereby reducing the number of trials needed and accelerating the design process.
Solution Approach 2:
The patent creates virtual replicas of material systems through computational models that simulate the behavior and properties of actual materials. These digital twins or virtual models allow for extensive testing and optimization in silico before physical experimentation, reducing the need for numerous physical trials while maintaining accuracy in property prediction.
3Manufacturing precision
If multiple manufacturing process routes are considered for material design, then the final material properties can be optimized, but the design process complexity increases significantly
Solution Approach 1:
The patent segments the complex multi-route manufacturing design problem into distinct modular components: composition optimization, process parameter optimization, structure prediction, and property evaluation. Each segment is handled by specialized sub-models that can be independently trained and validated, reducing the overall complexity while maintaining the ability to evaluate multiple process routes and their interactions.
4Ease of operation
If conventional design techniques rely on designer experience and known processing-structure-property relations, then the design process is straightforward, but the method lacks automation and integration of multiple process steps
Solution Approach 1:
The patent implements automated feedback loops where the system continuously evaluates predicted material properties against target specifications and automatically adjusts composition and process parameters accordingly. The integrated framework uses iterative optimization algorithms that learn from previous design attempts and automatically refine predictions, providing both automation and improved integration of multiple process steps while maintaining ease of operation through user-friendly interfaces.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
The disclosure relates generally to methods and systems for automated design of materials and the manufacturing process for desired properties. Conventional automated materials design techniques do not perform an integrated design of (i) a material composition and (ii) their manufacturing processing steps. The present disclosure addresses this gap by using a multi-agent setup for automated design, wherein a distinct Reinforcement learning (RL) agent is used to mirror the composition selection (CS) and various sequential manufacturing process steps (PS) involved in its manufacturing route. The distinct RL agents learn from both past design data and computational models (empirical/analytical/physics-based models) representing the design process. The present disclosure also integrates other important parameters such as manufacturability, ESG norms, cost, process energy etc. and their relative importance into the design decision making process of the RL agents by expressing them as reward components upon which the RL agents are trained.