基于手机APP传感器数据的个体出行特征精细化提取与交通模型优化

Nowadays, urban traffic problem is becoming increasingly serious and complex, traditional travel survey method with fuzzy, subjective and extensive data seems no longer adequate to satisfy planners' need in various aspects. Fortunately, the evolution of the new generation of wireless communications technology has brought new opportunities to solve these problems. In this project, a smart-phone application will be developed to collect the built-in smart-phone sensor data, such as GPS, acceleration and WiFi data; Wireless communication data collected from the 3G/4G-LTE system, for example handoff data, location update information, SMS, will be used to track cell-phone base station location changing trajectory; Intelligent data mining algorithms like wavelet analysis, random forests and cluster analysis will be explored to detect trip end, conduct trip segment division and detect travel mode; A ‘person-vehicle-communication’ integrated simulation platform will also be constructed to access the performance of accuracy and sensitivity of the proposed method. Finally, dominated by the smart-phone sensor data, fusing with the new generation of mobile communication data, an innovative system of individual travel trajectory data extraction methods will be fully introduced. Based on the system, the project will further try to study the theory and framework of the refined trip data based traffic model, research contents include traffic zone subdividing, calculation results improvement, traditional four-step model optimization, public transit OD estimation model assessment, activity-based travel model calibration and so on.

传统居民出行调查缺陷导致的模糊主观粗放式的数据质量瓶颈难以满足新时期复杂交通问题的分析需求，有待更精细高质数据支撑，新一代通信技术带来契机。研发智能手机APP提取内置传感器数据（GPS、加速度、WiFi等），解析3G/4G-LTE通信网络信令追踪手机用户在基站小区层级位置变化特征（软切换、短信、微信QQ上网等），以新型手机数据源为基础，运用小波分析、随机森林、聚类分析等数据挖掘算法，通过出行端点判定、出行轨迹切割和出行方式识别三个关键环节，耦合建立“人-车-通信”集成仿真平台辅助技术评估与敏感性分析，从而构建手机传感器数据为主导、融合新一代移动通信信令数据的个体出行活动特征精细化提取新方法体系，进而研究精细化出行数据对交通模型和理论的优化重构，包括从交通小区细分、数据质量改善、真实出行路径与交通分配等方面研究对四阶段模型精度提升，评估公交OD反推模型，实证标定基于活动的交通需求分析模型等。

传统居民出行调查缺陷导致的模糊主观粗放式的数据质量瓶颈难以满足新时期复杂交通问题的分析需求，有待更精细高质数据支撑，新一代通信技术带来契机。本项目研发智能手机APP提取内置传感器数据（GPS、加速度、WiFi等），解析3G/4G-LTE通信网络信令追踪手机用户在基站小区层级位置变化特征（软切换、短信、微信QQ上网等）。以新型手机数据源为基础，应用时空聚类、小波分析、机器学习等多种数据挖掘算法，从出行端点判定、出行轨迹切割和出行方式识别三个关键环节，构建了一套手机传感器数据为主导、融合新一代移动通信信令数据的个体出行活动特征精细化提取新方法体系。本项目通过产学研合作获取实证数据，完成技术在客观环境下的实证效果评估，并且耦合建立“人-车-通信”一体化集成仿真平台，完成技术在不同环境下的敏感性评估。在此基础上，本项目探索了利用手机数据提升或改进既有交通模型的可能性，以基于手机数据的交通小区划分尺度和优化方法为关键问题开展了研究，为不同基站密度下的交通小区划分尺度提供科学的依据。