Large Language Model–Enabled Decision Support for Complex Project Management

The rapid development of intelligent urban traffic systems has transformed traffic planning into a complex project management problem characterized by multi-stage decision-making, uncertainty, and strict policy constraints. Although large language models (LLMs) have recently demonstrated strong reasoning and planning capabilities, existing approaches often lack robustness, regulatory compliance, and long-term decision stability when applied to real-world urban traffic planning scenarios. To address these challenges, this paper proposes PM-LLM, a lifecycle-aware and tool-augmented LLM framework designed to support complex project management decision-making in intelligent urban traffic planning. PM-LLM integrates constraint-aware knowledge retrieval, uncertainty scenario modeling, hierarchical decision decomposition, simulation-grounded risk-sensitive selection, and closed-loop revision into a unified Plan–Simulate–Operate–Revise decision loop.