基于手机传感器数据的道路拥挤程度分析

发布时间：2018-01-05 15:22

本文关键词：基于手机传感器数据的道路拥挤程度分析　出处：《西南交通大学》2015年硕士论文　论文类型：学位论文

【摘要】：城市的高速发展给人们的生活带来了极大的便利和改善,然而城市发展同时带来了交通需求的不断增加,很多城市每天都发生着道路拥挤的情况。现如今,道路拥挤的问题已经成为了阻碍现代社会发展的一个关键问题。因此,快速识别城市道路的拥挤程度对改善交通出行、及时治理道路拥堵、保障人们的出行安全具有重要作用。另一方面,近几年智能手机飞跃式的进步使其成为智能交通研究的一个新方向,将智能手机中传感器的数据应用到道路拥挤程度识别的研究中,在实际应用中具有比较重要的价值。文本首先介绍了道路拥挤程度识别的国内外研究现状,并概述了城市道路拥挤的相关基础理论,再对目前应用的道路交通信息采集技术进行了总结。在研究了道路拥挤程度识别的原理、方法以及相关技术和算法的基础上,提出应用智能手机传感器的数据对道路拥挤程度进行识别,并对SVM的检测算法进行了研究。在实际识别的实验中,在分析采集到的传感器数据的基础上,确定输入模型的变量。对线性核函数、多项式核函数和RBF核函数下分类准确率的对比,在使用RBF核函数的情况下,利用交叉检验的方法择优寻求最优参数。根据以上的变量和参数综合对采集到的数据进行实验。显示RBF核函数取得了比较理想的检测准确率。本文还对比了不同数据处理方式下数据检测准确率的优劣,并根据实验结果分析了智能手机传感器数据在SVM算法下对道路拥挤程度识别的实际应用价值。
[Abstract]:The rapid development of the city has brought great convenience and improvement to people's life. However, the development of the city has also brought the increasing demand for traffic. Many cities have the situation of road congestion every day. Nowadays. The problem of road congestion has become a key problem that hinders the development of modern society. On the other hand, the advance of smart phone in recent years has made it a new direction of intelligent transportation research. The application of sensor data in smart phone to the research of road congestion recognition has important value in practical application. Firstly, the paper introduces the research status of road congestion recognition at home and abroad. And summarized the basic theory of urban road congestion, and then summarized the current application of road traffic information collection technology. In the study of the principle of road congestion recognition. Methods based on the related technologies and algorithms, this paper proposes to use the data of smart phone sensor to identify the road congestion, and studies the detection algorithm of SVM. On the basis of analyzing the sensor data collected, the variables of the input model are determined, and the classification accuracy of linear kernel function, polynomial kernel function and RBF kernel function are compared. In the case of a RBF kernel function. The method of cross test is used to find the optimal parameters. According to the above variables and parameters, the collected data are experimented. It is shown that the RBF kernel function has achieved a better detection accuracy. The accuracy of data detection under different data processing methods. Based on the experimental results, the practical application value of smart phone sensor data in road congestion recognition under SVM algorithm is analyzed.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495

【参考文献】