基于 AI 大模型的旅游公交线路动态优化研究
Research on Dynamic Optimization of Tourism Bus Route Based on AI Large Model
伴随智慧文旅的快速发展,旅游公交系统面对客流时空分布非稳态性强、多源信息耦合繁复及实时响应要求高等考验,传 统调度形式在准确性跟灵活性上已显不足。为此,本文建立了一种依据AI大模型驱动的动态改良框架,融合语义理解、跨 模态感知与轻量化部署技术,借助意图识别将非结构化游客需求映射为可计算向量,并融合边缘-云端协同架构实现毫秒级 异常检测与分钟级方略重生成。针对分布偏移跟计算瓶颈,引入滑动窗口在线学习跟双环反馈机制,在保障安全调度边界 的前提下加强模型适应性。实证说明,该方法明显减少等待时间与满载波动,增强系统鲁棒性与决策可解释性,为高动态 环境下公共交通智能调度给予了可行途径。
With the rapid development of smart cultural tourism, tourist bus systems face challenges such as highly non-stationary spatiotemporal passenger flows, complex multi-source information coupling, and stringent real-time response requirements. Traditional scheduling methods have shown limitations in both accuracy and flexibility. To address these issues, this study proposes a dynamic improvement framework driven by AI large models, integrating semantic understanding, cross-modal perception, and lightweight deployment techniques. By leveraging intent recognition, unstructured tourist demands are mapped into computable vectors, while an edge-cloud collaborative architecture enables millisecond-level anomaly detection and minute-level strategy regeneration. To tackle distribution shifts and computational bottlenecks, sliding-window online learning and a dual-loop feedback mechanism are introduced to enhance model adaptability while maintaining safe scheduling boundaries. Empirical results demonstrate that this approach significantly reduces waiting times and load fluctuations, improves system robustness, and enhances decision interpretability, offering a viable solution for intelligent public transport scheduling in highly dynamic environments.
| [1] |
钱宗强, 陈育军. 公交错时发车调度理论方法及应用[J]. 城市公共交通, 2025,(12):36-40. |
| [2] |
王晓彬. 定制公交战略维度思考[J]. 城市公共交通, 2025,(12):22-23. |
| [3] |
徐雷, 张国栋, 董文韬, |
| [4] |
赵光辉, 田园, 王雅婷, |
/
| 〈 |
|
〉 |