面向节假日客流波动的旅游公交AI优化策略研究

Research on AI Optimization Strategy for Tourism Buses Facing Holiday Passenger Flow Fluctuations

摘要

节假日期间旅游公交面对客流时空分布高度不均与运力刚性配置之间的突出矛盾，传统调度形式因依赖历史经验跟静态排班而不容易响应脉冲式需求波动，致使服务能力下降与乘客满意度减少。为加强系统解决极端客流的适应能力，本研究建立融合多源数据感知、深度学习预测跟强化学习调度的AI改良框架，在此基础上，引入多智能体强化学习机制，在分钟级粒度上改良发车间隔跟车辆配置，形成闭环弹性调度决策引擎；同时设计人机协同交互架构，集成调度员经验判断与算法推荐，增强系统鲁棒性跟可操作性。实验说明，该方案明显增强了高峰期运力匹配度跟运行稳定性，为智慧公交在繁复节假日场景下的智能化转型给予了可行途径。

Abstract

During holiday periods, tourist bus systems face a prominent contradiction between highly uneven spatiotemporal passenger flow distribution and rigid capacity allocation. Traditional scheduling approaches, which rely on historical experience and static timetables, struggle to respond to pulse-like demand fluctuations, resulting in reduced service capacity and lower passenger satisfaction. To enhance the system’s adaptability to extreme passenger flows, this study develops an AI-enhanced framework integrating multi-source data perception, deep learning-based forecasting, and reinforcement learning-driven scheduling. On this basis, a multi-agent reinforcement learning mechanism is introduced to optimize departure intervals and vehicle allocation at a minute-level granularity, thereby forming a closed-loop, flexible scheduling decision engine. Meanwhile, a human-machine collaborative interaction architecture is designed to incorporate dispatcher expertise with algorithmic recommendations, improving system robustness and operational feasibility. Experimental results demonstrate that the proposed approach significantly enhances capacity-demand matching and operational stability during peak periods, providing a feasible pathway for the intelligent transformation of smart public transport systems under complex holiday scenarios.

关键词

节假日客流预测 / 人工智能优化调度 / 旅游公交系统 / 图神经网络 / 强化学习

Key words

holiday passenger flow forecasting / AI-based scheduling optimization / tourist bus system / graph neural networks / reinforcement learning

参考文献

[1] 何莉, 刘健国, 陈鹏. 贵阳市城市公共交通发展问题及对策研究[J]. 综合运输, 2025, 47(12):159-163.

[2] 钟章建, 钱林波, 王天宇, 等. 宁波市主城区地面公交线网优化策略研究[J]. 交通与运输, 2025, 41(06):75-78.

[3] 张馨方, 闫艺萍, 张哲, 等. 基于车辆动力学和改进的FFRLS算法在线估算电动公交能耗[J]. 汽车安全与节能学报, 2025, 16(05):747-756.

[4] 王雅, 任静梅, 王妍文, 等. 基于统计与机器学习融合模型的水墨丹霞客流量预测优化研究[J]. 兰州交通大学学报, 2025, 44(06):63-72.

[5] Kim, Myung Ja, et al. ‘Does using public transport affect tourist subject well-being and behaviour relevant to sustainability? Value-attitude-behaviour theory and artificial intelligence benefits.’ Current Issues in Tourism 27.10 (2024): 1666-1682.