Back to Search
Dossier
基于深度学习的安全生产智能体交互与动作流执行方法
InventionUnknown
10Claims · 2 independent
§ Ⅰ
Dossier Overview
Inventor
张慧剑; 郎海亮
Filing Date
Aug 26, 2026
Publication Date
Oct 20, 2026
IPC Classification
G06N 3/006 (2023.01)G06N 3/042 (2023.01)G06N 3/045 (2023.01)G06N 3/048 (2023.01)G06N 3/084 (2023.01)G06N 3/092 (2023.01)
本发明涉及工业安全生产监控技术领域,公开了一种基于深度学习的安全生产智能体交互与动作流执行方法,包括:获取隐患时空状态描述向量和生产区域空间分布数据集;计算区域威胁清单;按预估到达时间划分梯度时间区间并生成梯度预加载调度计划;利用图注意力网络生成跨分支数据耦合关系图谱;并行执行动作流分支并识别修正需求;通过多智能体协作强化学习架构生成协同修正策略组合;传播约束更新至关联分支;执行跨分支一致性校验;动态更新调度计划并生成处置报告。
Sign in to read the full text
Claim 1 is shown. Create a free account to read all 10 claims, the full description, all drawings, translations and legal status.
- All claims
- Technical field
- Background
- Summary
- Drawing descriptions
- Detailed description
External Resources