基于WiFi定位数据的室内群体活动演化过程建模方法

Indoor space is an important place for human activities. With the development of cities, increasingly complex indoor space has brought great challenges to indoor public safety, emergency management and spatial planning. Indoor human activities are deeply influenced by indoor spatial structure, which makes it a key to reveal the law of interaction between human and indoor environment and solve the challenge of indoor spatial applications through comprehensive understanding of indoor human activities’ spatial-temporal patterns. How to identify the characteristics of indoor human activities from massive WiFi positioning data and establish a spatial-temporal unified model to describe the evolution patterns of indoor human activities are two key scientific quests in this project. Starting from the analysis of the effect of indoor spatial structure on the positioning accuracy of activity observations, this project plans to investigate an error correction method of the large-scale WiFi positioning data from a real-world application, design an inference algorithm of individual activity type that takes into account the indoor environmental constraints, and developed a process-oriented model of pattern evolution process of indoor human activities’ multi-dimensional feature. To break through the traditional macro-scale-oriented, distance-based modeling methods of activity context and spatial-temporal segmented analysis in outdoor human activity research, so as to develop an indoor-oriented, process-oriented research method of indoor human activities. It has important research and social significance to explore the evolution pattern of indoor human activities and guide the reasonable planning and accurate management of complex indoor space.

室内空间是人类活动的重要场所。随着城市发展，日益复杂的室内空间给室内公共安全、应急管理、空间规划带来巨大挑战。室内人群活动受室内空间结构的深刻影响，对其时空模式的全面理解已成为揭示人与室内环境交互作用规律、解决室内空间应用难题的关键。如何从海量WiFi定位数据中识别室内群体活动特征、建立时空统一的模型挖掘室内群体活动的特征演化规律是本项目研究中的两个关键科学问题。本项目以室内空间结构对人群活动观测精度影响分析为切入点，研究大规模WiFi定位数据误差校正方法，设计顾及室内环境约束的个体活动属性推断算法，构建面向过程的室内群体活动多维特征演化过程模型。突破传统人群活动研究中面向宏观尺度、以距离为主的活动上下文建模方法、时间和空间分割的分析模式，面向室内发展顾及空间几何特征、时空统一的人群活动分析方法。对探索室内人群活动演化规律，指导室内空间的科学合理规划与精准管理具有重要的研究和社会意义。

人群活动受室内外环境空间结构的深刻影响，对其时空模式的全面理解已成为揭示人与室内外环境交互作用规律、开发空间应用的关键。项目围绕室内WiFi定位数据等泛在定位数据支持下的人群活动演化过程建模，重点开展了三个方面研究：针对泛在定位数据中的位置噪声，按照地图辅助轨迹纠偏的总体思路，对泛在地图信息解译、地图辅助轨迹纠偏算法、线要素精度分析的技术全流程开展研究，并在大型室内工程项目中取得应用；面向个体活动属性获取，以环境空间结构特征作为研究对象，开展面向个体行为理解的空间结构认知研究，尝试从认知机理的角度，揭示环境空间结构对个体空间行为的影响作用；基于位置纠偏、语义完整的泛在定位数据，以活动热点为线索，从时空特征动态变化的角度，研究了人群活动进行建模与分析技术，探讨了移动网络结构下的群体活动演变规律。