基于BP神经网络的网约车出行需求短时预测

发布时间：2018-05-17 00:48

本文选题：网约车 + 供需匹配度　；参考：《北京交通大学》2017年硕士论文

【摘要】：随着“互联网+”在各行业的不断渗透,为传统行业的革新与升级注入了全新的动力。巡游出租车行业供需时空信息的不对称性导致“打车难”问题持续存在,使其成为“互联网+”必然会触及的领域。由此催生了网约车这一新业态,有效打通了出租车供给与需求之间的信息不对称性问题。本文利用网约车平台的公开数据,开展了网约车需求特性分析与短时预测,为网约车运营提供方法性参考,对提升供需匹配效率具有重要意义。本文主要工作如下:首先,本文通过查阅大量研究文献,从传统出租车供需、互联网时代的出租车供需研究、以及交通短时预测等方面进行了综述,梳理了本文研究内容与技术路线。并且通过对比网约车与巡游出租车的行业特性,进一步明确了本文的研究内容。第二,网约车出行需求时空特性分析。根据网约车出行数据的特征,本文将网约车出行需求总数拆分为供需匹配数与需求缺口数,并定义了网约车供需匹配度与需求紧缺度。并据此分析工作日与非工作日网约车出行需求的时间特性,划分了不同的时段类型,并得出工作日与非工作日供需匹配度的差异性。然后在此基础上进行分时段的网约车出行需求空间特性分析,为网约车出行需求短时预测提供了依据。第三,网约车出行需求短时预测。论文从现实意义角度出发,以需求缺口数作为网约车出行需求短时预测的目标,并进行了时间相关性分析,发现网约车出行需求缺口与历史前50分钟,以及同时刻历史日期的出行需求缺口相关程度较大。根据网约车出行需求缺口的特点构建了基于BP神经网络的网约车出行需求短时预测模型,依据相关性分析结果确定了模型结构,并以网约车出行实际数据进行了需求短时预测并验证了模型有效性。最后,给出了改善网约车供需匹配的相关建议,并对下一步研究进行了展望。
[Abstract]:With the continuous penetration of the Internet in various industries, the innovation and upgrading of traditional industries has injected a new impetus. The asymmetry of time and space information between supply and demand of tour taxi industry leads to the persistence of the problem of "taxi hailing difficulty", which makes it an inevitable area to be touched by the "Internet". This leads to the birth of a new form of cable-sharing, which effectively solves the problem of information asymmetry between the supply and demand of taxis. This paper makes use of the open data of the network car-hailing platform to carry out the analysis of the demand characteristics and the short-term forecast of the network-ride-hailing demand, which provides a methodological reference for the network-ride-hailing operation and is of great significance to the improvement of the efficiency of the matching between supply and demand. The main work of this paper is as follows: firstly, this paper reviews the traditional taxi supply and demand, the research of taxi supply and demand in the Internet era, and the short-term traffic forecasting through consulting a large number of research documents. Combing the research content and technical route of this paper. And by contrasting the industry characteristic of the net and tour taxi, the research content of this paper is further clarified. Second, the time-space characteristic analysis of the travel demand. According to the characteristics of the trip data, this paper divides the total demand into the supply and demand matching number and the demand gap number, and defines the supply and demand matching degree and the demand shortage degree. Based on the analysis of the time characteristics of the demand for car-sharing between workday and non-workday, different time periods are divided, and the difference of supply and demand matching degree between workday and non-workday is obtained. On the basis of this, the spatial characteristics of the travel demand are analyzed, which provides the basis for the short-term forecast of the demand for the net-ride-hailing trip. Third, the net car ride demand short-term forecast. From the point of view of practical significance, this paper takes the number of demand gap as the goal of short-term forecast of the demand for car-hailing travel, and analyzes the correlation of time, and finds that the gap of demand for network-ride-sharing travel is 50 minutes before history. And at the same time the historical date of travel demand gap is relatively large. According to the characteristics of the gap in the demand for ride-to-vehicle travel, a short-time forecasting model of the demand for ride-hailing trip is constructed based on BP neural network, and the structure of the model is determined according to the results of correlation analysis. Based on the actual data, the demand is predicted in a short time and the validity of the model is verified. Finally, some suggestions to improve the matching between supply and demand are given, and the future research is prospected.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;U491

【参考文献】