数据挖掘技术在高速公路联网收费稽查上的研究与应用

发布时间：2018-12-16 01:01

【摘要】：随着高速公路联网收费规模不断扩大，通车里程不断增加，单次高额通行费诱惑导致逃费作弊行为愈演愈烈，偷逃通行费手段不断翻新，一方面造成国家规费的流失，另一方面造成缴费不公，扰乱交通运输市场秩序，也给高速公路运营管理工作带来很大的难度。常态稽查以高速公路联网收费后台数据综合分析为基本手段，通过人工手段从高速公路联网收费海量数据库开展稽查分析，效率极为低下，漏判、误判重要逃费线索数据常有发生。运用数据挖掘技术手段对当前联网收费系统的数据进行处理，充分挖掘数据中的信息价值，提高工作效率，利用科技手段有效防范和打击偷漏通行费行为，是高速公路联网收费稽查工作开展的突破口和着力点，同时建立健全高速公路联网收费稽查工作的长效机制，实现稽查手段由人工判别向科技辅助转变，促进高速公路联网收费稽查管理信息化建设具有重要意义。本论文是在对高速公路联网收费系统数据库中部分值得研究的可疑数据进行清洗，建立相相应的数据仓库，并通过对数据仓库中信息的分析和数据挖掘，提取有价值或者以往没有发现的信息，满足高速公路运营管理单位收费稽查工作中的数据分析决策需要。以广东省高速公路联网收费系统为代表阐述了高速公路联网收费系统结构、组成及相关原则和联网收费数据模型，，根据高速公路联网收费各运营路段防逃费的调研总结，对逃费行为进行了深入研究与分析，提取其异常行为的共有特征，利用SQLServer2008R2先进的数据挖掘功能，以广东省粤东片区某路段2012年的历史数据为挖掘对象，选取决策树算法预测OD定向时间进行动态超时设置和关联规则算法挖掘出团伙性逃费嫌疑的异常流水数据。根据挖掘的结果，再结合路段收费稽查管理系统，可提取原始交易流水与相应的佐证信息（如出入口图像、高清卡口图像、过车视频等），为路段运营部门判别逃费车辆和防止逃费发生提供参考依据。
[Abstract]:With the continuous expansion of toll scale and mileage of highway network, the single high toll temptation leads to more and more cheating, and the means of evading tolls continue to be renovated, which on the one hand results in the loss of national fees. On the other hand, it causes unfair payment, disrupts the order of transportation market, and brings great difficulty to highway operation and management. The normal inspection is based on the comprehensive analysis of the backstage data of the expressway network toll collection, and carries out the audit analysis from the massive database of highway network toll collection by artificial means. The efficiency is extremely low and the judgment is missing. Miscalculation of important escape fee leads to data often occur. Using data mining technology to deal with the data of the current network toll collection system, fully mining the value of information in the data, improving work efficiency, using scientific and technological means to effectively prevent and crack down on the behavior of stealing tolls. It is the breakthrough and focus of the highway network toll inspection work. At the same time, it is necessary to establish and improve the long-term mechanism for the expressway network charge inspection work, and to realize the transformation of the inspection means from manual discrimination to science and technology assistance. It is of great significance to promote the information construction of highway network charge checking and management. In this paper, some suspicious data worth studying in the database of expressway network toll collection system are cleaned, the corresponding data warehouse is established, and the information in the data warehouse is analyzed and mined. The valuable or undiscovered information can be extracted to meet the needs of data analysis and decision in toll audit of expressway operation and management units. Taking Guangdong Provincial Expressway Network Toll system as the representative, this paper expounds the structure, composition, related principles and data model of the Expressway Networked toll collection system, and summarizes the investigation and summary of the anti-escape charges in each operating section of the Expressway Network Toll Collection. In this paper, the behavior of fee evasion is deeply studied and analyzed, and the common characteristics of abnormal behavior are extracted. Using the advanced data mining function of SQLServer2008R2, the historical data of a section of Guangdong province in 2012 is taken as the mining object. The decision tree algorithm is selected to predict the OD orientation time for dynamic timeout setting and association rule algorithm to mine the abnormal pipeline data. According to the results of mining, combined with the section charge inspection management system, the original transaction flow and corresponding supporting information (such as entrance and exit image, high-definition bayonet image, passing video, etc.) can be extracted. It provides the reference for the section operation department to distinguish the evading vehicle and to prevent the evading fee.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;TP311.13

【参考文献】