基于神经网络的沥青混合料配合比设计  

Design on Asphalt Mixture Ratio Based on Neural Networks

作　　者：孙益民[1] 钟明[1] 徐玲[1] 蒋伟华 范凯 

机构地区：[1]安徽师范大学化学与材料科学学院,安徽芜湖241000 [2]中铁四局集团公司,安徽合肥2300012 [3]安徽省技术创新服务中心,安徽合肥230001

出　　处：《公路交通科技》2012年第2期40-45,共6页Journal of Highway and Transportation Research and Development

基　　金：安徽省教育厅自然科学基金重点项目(KJ2011A149);科技部科技型中小企业技术创新基金(11C26213401894)

摘　　要：为了改良现有混料理论不能很好解决混合料的各种力学性能优化的问题。通过设计数十种不同配合比试验,分别测试按不同配合比制备的沥青混合料的主要性能指标。运用人工神经网络分析方法,以沥青结合料、水泥和4种不同粒径的矿石的用量比例为模型的输入,以沥青混合料的马歇尔稳定度和流值为输出,建立6-4-2的BP神经网络模型,对混合料的马歇尔稳定度和流值两大性能指标进行了研究。从而对沥青结合料、水泥和4种不同粒径的矿石制备沥青混合料的工艺进行了优化,并找到了理想马歇尔稳定度和流值时的适宜配合比为:C1>32%、16%<C2<40%、C3<22%、12%<C4<36%、3.5%<C<7.5%、O/S>7%。In order to solve the problem that existing mixing theory cannot optimize various mechanical properties of asphalt mixtures, dozens of experiments in different mixture ratios were designed to test the main performance indexes of each experiment. The Marshall stability and flow values of the mixtures were studied with artificial neural network and the established 6-4-2 BP neural network model, the amount of asphalt binder, cement and four different sizes of ore were inputs, Marshall stability and flow values were outputs. The producing technology of asphalt mixture which mixed with asphalt binder, cement and four different sizes of ore was optimized. Finally, the better mixture ratio in ideal Marshall stability and flow values was obtained： C1 〉32%, 16% 〈C2 〈40, C3 〈22%, 12% 〈C4 〈36%, 3.5% 〈C〈7.5%, and O/S〉7%.

关 键 词：道路工程 适宜配合比 神经网络 沥青混合料 马歇尔稳定度 流值 

分 类 号：U416.217.02[交通运输工程—道路与铁道工程]