Large model material formula design and evaluation method and system based on multi-agent collaboration and storage medium
By employing a large-scale model approach based on multi-agent collaboration, the problem of lack of data governance and mechanism constraints in material formulation design is solved, enabling efficient and accurate formulation generation and evaluation, and filling the application gap of multi-agent systems in material formulation design scenarios.
Patent Information
- Application Number
- CN202511408661.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-29
- Publication Date
- 2025-10-31
- Estimated Expiration
- 2045-09-29
AI Technical Summary
Existing material formulation development relies on experimental trial and error and experience, which is time-consuming and costly. A single large model cannot simultaneously complete multi-source data cleaning and quality control. The lack of mechanistic constraints leads to unreliable generated formulations. Multi-agent systems lack full-process integration in material formulation design.
A large-scale model approach with multi-agent collaboration is adopted. Data agents clean multi-source data, generating agents generate initial formulas, mechanism agents screen compliant formulas, evaluation agents perform performance estimation and risk assessment, and decision agents output the optimal formula. By combining an explicit mechanism knowledge base and a lightweight prediction model, a closed-loop collaboration between data governance and model generation is achieved.
It improves the efficiency and accuracy of material formulation design, reduces invalid experimental verification, ensures that the formulation meets environmental protection standards and performance requirements, supports pluggable large model adaptation, and improves R&D efficiency.