【作者】 陈刚；

【导师】 李勇；

【作者基本信息】 重庆大学 ， 图书馆、情报与档案管理， 2014， 硕士

【摘要】 在计算机、互联网以及移动通信出现以后，物联网引领了新一代的信息技术革命。2009年美国在首次提出“智慧地球”这一新颖概念的同时，将物联网作为振兴经济、提高国家实力的重点领域之一。同一年，我国温总理提出了“感知中国”，并把物联网正式列入国家五大新兴战略性支柱产业之一，同时写入政府工作报告。在信息社会，物联网主要应用在医疗、电子政务、电网、教育、交通、城市管理等领域。物联网应用的同时产生了数以万计的实时数据，随着数据量的急剧增长和数据种类的不断增加，催生了大数据时代的到来。车联网作为物联网在智能公交中的典型应用，车联网条件下的车载数据除了具有物联网大数据一般性的特征，而且具有自身的结构性质。本文在总结数据预处理、数据清洗和数据质量等情报定量分析的相关文献下，探讨了实时数据清洗处理定量分析的研究现状，分析了车联网实时数据的主要质量问题和特征。在研究一般传感数据和传感信息的处理方法下，基于滑动窗口借鉴使用各种情报定量分析方法对车载传感实时数据进行清洗等预处理。针对所研究的问题，本文介绍了车联网网络架构和车载数据框架，结合车载数据的海量、异构、实时等主要特征，提出了车联网条件下的车载数据实时清洗的预处理框架，然后以连续型和开关量信号两种主要的车载类型实时数据的清洗处理算法。最后以城市混合动力客车的天然气气瓶压力为代表的连续型变量和刹车制动为代表的车载开关量信号实时数据为基础，分别研究分析了相关的实时清洗预处理算法。针对各自的数据特征，本文以滑动窗口来保证数据清洗的实时性，运用莱茵达准则、滑动平均滤波、小波变换、贝叶斯决策理论等多种定量分析方法进行混合清洗处理使用。运用这些情报定量分析方法对数据进行预处理，能够有效地处理掉错误、噪声、缺失等异常数据，得到准确的数据信息，为驾乘人员提供实时、准确、可靠的公共交通信息服务，从而改善公共交通的智能化运营管理。由于车载数据的特征和数据实时清洗算法的复杂性，加之写作时间比较仓促和自身的水平能力有限，文中有错误之处还请批评指正。更多还原

【Abstract】 After following the computer, Internet and mobile communication network, theInternet of things has led to a new generation of information technology revolution. In2009first proposed the innovative concept of "wisdom of the earth", and at the same time,the Internet of things as the revitalization of the economy, one of the key areas to improvenational strength. The same year, Chinese Premier Wen put forward the "perceptionChina" since the Internet of things has been officially listed as one of the five majoremerging strategic industry in our country, and written into the government work report.In the information society, the Internet of things is mainly applied in the field of medical,electronic government affairs, network, education, transportation, city management. TheInternet of things applications also produced tens of thousands of data, with the growingvolume of data increases dramatically and data types, and gave birth to the arrival of theera of big data.Car networking as the Internet of things in the typical applications of intelligentpublic transportation, in addition to the General characteristics of the Internet data, butalso has its own structures and properties of vehicle data under the condition oftelematics. In this paper, under related documentation of data quality and data cleaningand data pre-processing and such as quantitative analysis, discussed the current researchstatus of real-time data cleaning process quantitative analysis, combined with analysis ofthe main quality problems and characteristics of networking real-time data. In theresearch general sensing data and under the sensing information processing method, usinga variety of intelligence quantitative analysis method of data cleaning pretreatment ofon-board sensors based on the hybrid window.For the study of the issues raised, proposed the telematics network construction andon-board data frame, combined with the massive, heterogeneous, real-time and theprincipal features of vehicle data, proposes telematics data pre-processing framework ofreal-time cleaning, which is based on a continuous-type and switch signal from two maintypes of on-board data that in real-time cleaning algorithm. Finally, with the gas cylinderpressure as the representative of the continuous variables and the brake of on-boardswitch signal real-time data as the foundation, and were studied and analyzed the relevantalgorithm of real-time cleaning. According to the data of respective characteristic, basedon the sliding window to ensure real-time data cleaning, use the lander criterion, moving average filter, wavelet transform, Bias decision theory and many kinds of quantitativeanalysis methods for fusion cleaning. Use these intelligence quantitative analysis methodsfor data preprocessing, and to be able to effectively deal with the error, noise, missing,such as abnormal data deletion and obtains the accurate data message. And providingreal-time, accurate and reliable public transport information service for the occupants andso as to improve intelligent management of public transport.Due to the complexity of real-time data cleaning algorithm and the characteristics ofvehicle data, coupled with the writing time is quite hasty and own horizontal ability islimited, there are places of the mistakes and also invited to criticize and point outmistakes in the article.更多还原