基于决策树的沥青路面使用性能评价与预测研究  

Study on the Evaluation and Prediction of Asphalt Pavement Performance Based on the Decision Tree

作　　者：纪红燕 王钢 俞长青 吴启勋[2] 

机构地区：[1]青海弘川新源实业股份有限公司,西宁市810007 [2]青海民族大学化学化工学院,西宁市810007

出　　处：《公路》2016年第4期18-22,共5页Highway

基　　金：青海省科技攻关项目;项目编号2012-N-129

摘　　要：准确的评价和预测可以为路面养护管理提供科学的决策依据。针对统计学方法、物元模型以及神经网络等在沥青路面使用性能评价方面存在的问题,提出了决策树模型。选取现行规范确定的沥青路面使用性能评价指标,以沈阳~大连高速公路路况数据为基础,采用数据挖掘SPSS Clementine软件,建立了高速公路沥青路面使用性能决策树评价模型。和文献已有的物元模型相比,分类树合理有效,而回归树评价与预测准确,具有特有的优越性。研究结果表明,决策树模型是可靠的,可作为评价和预测沥青路面性能的一种新实用方法。Accurate pavement performance evaluation and prediction can provide scientific decisionmarking basis for pavement maintenance management.Regarding the existing problems of statistical method,model of matter element and neural network for asphalt pavement performance evaluation,the decision tree model is presented.Several main evaluation indexes determined by current specifications are employed to evaluate the performance of asphalt pavement.The decision tree evaluation model of asphalt pavement with SPSS Clementine for data mining software is presented based on pavement performance data of Shenyang-Dalian Expressway.Compared with the results based on model of matter element,it could be indicated that the classification tree is feasible and effective,evaluation and prediction of the regression tree is accurate and show special advantages.This decision tree model is proved to be a reliable and new practical method for evaluation and prediction of asphalt pavement performance.

关 键 词：决策树 沥青路面 路面使用性能 评价与预测 SPSS CLEMENTINE 

分 类 号：U416.217[交通运输工程—道路与铁道工程]