轨迹驱动的滨海景区自驾游客流时空特征及分流模型研究

The spatial behavior and motion pattern of tourists is one of the core contents of tourism geography research. Since the self-driving tourists has become an important part of scenic tourists flow in recent years, it has caused difficult parking, scenic congestion, air pollution and other issues. Firstly, in order to effectively control and navigate self-driving tourist flow and ease the pressures of traffic facilities and environment in scenic area, this research focuses on the spatial behaviors of self-driving tourists in microscopic scale based on trajectory data and explores the analysis methods by utilizing the theories of traffic geography, time geography and behavior science. Secondly, we take coastal scenic as an example to study the movement pattern of tourists. Then, we will analyze the time-varying characteristics of trajectory data and quantify its influence on route selection and parking choice behavior. Finally, in order to guide the equilibrium distribution of tourists on space and time, we will propose a control modeling by considering the motion pattern, the capacity of scenic parking lots and the preferences of route selection. Through this research, we can accurately grasp the distribution status and evolution process of the self-driving tourists, and provide decision-making basis for scenic planning and congestion control.

旅游流的空间行为及运动模式是旅游地理学的核心内容之一。近年来自驾游者成为景区客流的重要组成部分，在旅游高峰期产生了停车难、景区拥堵、空气污染等问题。为了有效诱导及均衡分配自驾客流，缓解景区及周边的交通设施与环境压力，本项目以重构自驾客流时空轨迹为基础，结合交通地理学、时间地理学、行为科学等理论，探索微观尺度下景区自驾游客流空间行为分析方法；以研究区为案例，通过对客流的时空分布特征、演化过程以及流动模式研究，揭示滨海景区自驾游客流运动规律；引入轨迹信息中的时变属性，量化分析其对自驾游路线选择行为与停车选择行为的影响；综合景区客流运动模式、停车容量、线路选择偏好等构建时空分流模型，引导客流在景区实现时空均衡分配。通过本项目研究，可为景区旅游规划和建设、客流管理提供决策依据。

近年来自驾游客成为景区客流的重要组成部分，在旅游高峰期产生了停车难、景区拥堵、空气污染等问题。为了有效诱导及均衡分配自驾客流，缓解景区及周边的交通设施与环境压力，本项目以重构自驾客流时空轨迹为基础，结合交通地理学、时间地理学、行为科学等理论，探索微观尺度下景区自驾游客流运动规律、空间流动模式、时空分流建模方法。首先，本项目结合研究区兴趣点数据与社区检测提出了自驾客流监测优化布点算法，建立了微观尺度下的自驾客流数据集，分析了自驾游客流演化特征、空间分布特征。其次，根据景点间是否存在自驾游客流动构建起空间交互网络图集，并结合频繁图挖掘算法提取了景区自驾游客流动模式，对比分析了市内、城市间的自驾游客流动特点及路径选择偏好。鉴于高峰期景区自驾游停车路径受到停车场及附近路径的客流关联影响，提出了基于多头注意力与图神经网络的多监测点客流综合预测算法。最后，结合多智能体技术、节点分配模型、车辆跟驰模型，建立了基于多智能体的自驾游客流时空分流模型，并开发了相应的软件系统。通过本项目研究，可为景区旅游规划和建设、客流管理提供决策依据。