Machine-Learned Capability Maps for Material Property Achievability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
OEMs face challenges in efficiently co-optimizing product performance with materials specifications due to a lack of data on the difficulty of achieving specific material properties, leading to inefficiencies in material development and increased costs.
Innovation Solution
A system that generates a capability map using machine learning to predict material performance and likelihood of success, allowing OEMs to evaluate materials suppliers' capabilities without access to proprietary data, thereby optimizing product design and material selection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If OEMs request new material development from suppliers, then material performance specifications can be achieved, but development time and cost increase due to lack of data on achievability
Solution Approach 1:
The system performs preliminary analysis of material specification achievability using machine learning models trained on historical supplier data. Before initiating material development, the system evaluates the likelihood of success and provides feedback to OEMs, allowing them to adjust specifications or select alternative materials before committing resources to development.
Solution Approach 2:
The system implements a feedback mechanism where historical material development outcomes from suppliers are continuously fed into machine learning models. These models then provide predictive feedback to OEMs about the achievability of requested material specifications, enabling informed decision-making and reducing wasted development efforts on unachievable targets.
2Reliability
If OEMs request new material development from suppliers, then material performance specifications can be achieved, but development costs increase due to uncertain difficulty levels
Solution Approach 1:
The system performs preliminary cost assessment by analyzing historical data on material development difficulties and supplier capabilities. Before OEMs commit to material development, the system provides cost estimates based on the predicted difficulty level and supplier expertise, enabling budget planning and specification optimization.
Solution Approach 2:
The system uses feedback from historical material development projects to continuously improve cost prediction accuracy. By analyzing patterns in supplier performance across different material types and specification difficulties, the system provides increasingly accurate cost estimates that help OEMs optimize their material requests for cost-effectiveness.
3Loss of information
If OEMs lack data on supplier capabilities, then material specification difficulty cannot be assessed, but co-optimization of product performance with material specifications becomes inefficient
Solution Approach 1:
The system creates a virtual copy or digital twin of the supplier's material development capabilities by training machine learning models on historical supplier data. This digital representation allows OEMs to query and assess supplier capabilities without direct access to proprietary supplier information systems, enabling efficient co-optimization while preserving data security.
Solution Approach 2:
The system acts as an intermediary layer between OEMs and suppliers, translating supplier capability data into actionable insights for OEMs while maintaining data security. The machine learning models process and anonymize supplier data, providing OEMs with the information needed for efficient material specification setting without requiring direct access to supplier proprietary systems.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A device generates a capability map. The device receives one or more design spaces from a materials supplier, the one or more design spaces including candidate components and capabilities of tools available to the materials supplier. The device inputs a design space of the one or more design spaces into a machine learning model, the training data including a plurality of components including input materials and/or chemicals, and, for respective combinations of the plurality of components, a plurality of respective performance properties. The device receives as output from the model a capability map of the materials supplier storing possible combinations of performance properties and a respective difficulty of developing a composition with that combination of performance properties. The device outputs a user interface for display to a user indicating data of the capability map.