Entropy-Constrained Strategy Optimization in Urban Floods: A Multi-Agent Framework with LLM and Knowledge Graph Integration

2025-08-20

AI Summary

Urban flood emergency dispatch faces three core challenges: difficulty in multi-objective trade-off optimization, high environmental dynamism, and semantic instability of LLM-generated strategies—hindering unified real-time perception, global optimization, and multi-agent coordination. To address these, we propose H-J, a hierarchical multi-agent framework integrating large language models (LLMs) with domain-specific knowledge graphs. H-J introduces two novel mechanisms: entropy-constrained strategy generation to enhance output consistency, and knowledge-guided prompting for grounded, interpretable reasoning. It establishes a closed-loop optimization pipeline spanning real-time sensing, knowledge-augmented decision-making, feedback-driven refinement, and coordinated multi-agent control. Evaluated on real-world urban topography and rainfall datasets, H-J significantly outperforms rule-based baselines and reinforcement learning approaches across key metrics—including traffic fluidity, mission success rate, and system robustness—demonstrating the first unified framework that jointly achieves dynamic environment awareness, knowledge-enhanced inference, adaptive optimization, and scalable multi-agent coordination.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Multiagent Systems: Multiagent Planning
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

In recent years, the increasing frequency of extreme urban rainfall events has posed significant challenges to emergency scheduling systems. Urban flooding often leads to severe traffic congestion and service disruptions, threatening public safety and mobility. However, effective decision making remains hindered by three key challenges: (1) managing trade-offs among competing goals (e.g., traffic flow, task completion, and risk mitigation) requires dynamic, context-aware strategies; (2) rapidly evolving environmental conditions render static rules inadequate; and (3) LLM-generated strategies frequently suffer from semantic instability and execution inconsistency. Existing methods fail to align perception, global optimization, and multi-agent coordination within a unified framework. To tackle these challenges, we introduce H-J, a hierarchical multi-agent framework that integrates knowledge-guided prompting, entropy-constrained generation, and feedback-driven optimization. The framework establishes a closed-loop pipeline spanning from multi-source perception to strategic execution and continuous refinement. We evaluate H-J on real-world urban topology and rainfall data under three representative conditions: extreme rainfall, intermittent bursts, and daily light rain. Experiments show that H-J outperforms rule-based and reinforcement-learning baselines in traffic smoothness, task success rate, and system robustness. These findings highlight the promise of uncertainty-aware, knowledge-constrained LLM-based approaches for enhancing resilience in urban flood response.

Problem

• Optimizing emergency scheduling in urban floods with dynamic strategies
• Addressing semantic instability in LLM-generated flood response plans
• Integrating multi-agent coordination under rapidly changing rainfall conditions

Innovation

• Hierarchical multi-agent framework integration
• Knowledge-guided prompting and entropy-constrained generation
• Feedback-driven optimization with closed-loop pipeline