高速公路通行卡逃费行为预测模型研究

发布时间：2018-01-21 01:46

本文关键词： 交通工程高速公路通行卡逃费聚类分析判别分析逻辑回归　出处：《中国科技论文》2015年19期 　论文类型：期刊论文

【摘要】：针对当前利用通行卡进行高速公路逃费的现象较为严重而手工稽查效率低、效果差的问题,采用聚类判别分析框架研究高速公路逃费行为,运用逻辑回归分析方法探索高速公路通行卡(ETC)逃费概率的影响因素,提出了通行卡逃费行为的预测新模型。发现聚类判别分析框架对高速公路逃费通行卡的识别效率更高,且有利于提高命中能力;车辆属性、出行属性、超重属性、黑夜出行属性都是影响通行卡逃费概率的显著因素;经验证,预测新模型对通行卡逃费行为的预测正确率达到99.72%。结果表明:采用聚类分析、判别分析和逻辑回归分析相融合的方法,是识别和预测高速公路通行卡逃费行为的有效途径;预测新模型有利于管理部门制订科学的逃费管理解决方案,减少逃费稽查的成本和工作量。
[Abstract]:In view of the current use of traffic card highway evasion phenomenon is more serious and manual inspection efficiency is low, the problem of poor results, using clustering discriminant analysis framework of evasion behavior of highway, using logistic regression analysis to explore the highway toll card (ETC) factors "probability, put forward the new traffic prediction model of evasion behavior. Clustering card discriminant analysis framework on Higher Highway evasion pass card recognition efficiency, and is helpful to improve the shooting ability; vehicle attributes, travel attribute, overweight property, night travel attributes are significant factors affecting traffic card evasion probability; after verification, the prediction model of the new card evasion behavior on the passage of the results show that the correct rate of 99.72%. using cluster analysis, discriminant analysis and logistic regression analysis method of integration, is the identification and prediction of highway toll evasion line card Effective way to; a new prediction model is conducive to the management department to formulate scientific management "solution, reduce the cost and workload of evasion of inspection.

【作者单位】：云南省高速公路联网管理中心;华南理工大学电子与信息学院;广州华工信息软件有限公司;
【基金】：云南省交通运输厅科技专项资金资助项目(云交科2015(A)06)
【分类号】：U491
【正文快照】： 在全国高速公路联网收费即将实现的大背景下[1],由于利益的驱动,使得利用通行卡等手段进行车辆偷逃通行费(以下简称“逃费”)的作弊现象有愈演愈烈之势[2]。据统计,逃费流水占收费站出口流水总数的0.4%[3],逃费损失可达通行费总收入的5%[4],资料显示,每年我国高速公路通行费损

【参考文献】