CP-Agent:面向晶体塑性仿真工作流的工程化智能体
原标题:CP-Agent: A Harness-Engineered Agent for Crystal Plasticity Simulation Workflows
AI 摘要
研究提出 CP-Agent,一个基于 LLM 的 harness 工程化智能体,可在自然语言任务描述下自主执行完整的晶体塑性(CP)建模工作流。该智能体采用 ReAct 范式进行工具选择与排序推理,并将数值搜索交给已有优化器,通过类型化工具定义、调度器和安全受限迭代循环编码领域知识。在四个案例中(316L 增材制造不锈钢滑移参数标定、铜基准验证、铜初始织构恢复、Mg-Zn-Ca 合金多道次轧制织构演化),智能体均能从任务陈述推断正确执行顺序并给出物理可解释结果。
正文节选
[orcid=0009-0000-2142-0555] [orcid=0000-0002-1213-7958] CP-Agent: A Harness-Engineered Agent for Crystal Plasticity Simulation Workflows Abstract Crystal plasticity (CP) simulations are widely used to predict the mechanical behavior of polycrystalline metals, yet their routine application remains hindered by the substantial manual effort of configuring heterogeneous tools, orchestrating multi-step data pipelines, and calibrating constitutive parameters against experimental measurements. These bo