基于不同算法的乳化沥青冷再生材料养生性能预测  

Prediction of curing performance of cold recycled materials of emulsified asphalt based on different algorithms

作　　者：杨彦海 逄海洋 杨野 YANG Yan-hai;PANG Hai-yang;YANG Ye(School of Traffic Engineering, Shenyang Jianzhu University, Shenyang 110168, China)

机构地区：[1]沈阳建筑大学交通工程学院,沈阳110168

出　　处：《沈阳工业大学学报》2021年第4期455-462,共8页Journal of Shenyang University of Technology

基　　金：国家自然科学基金项目(51178278);辽宁省自然科学基金项目(20162631).

摘　　要：为了揭示乳化沥青冷再生混合料养生条件与养生性能之间的内在联系,利用不同算法对室内成型的乳化沥青冷再生试件的养生性能建立预测模型.以养生条件作为模型输入参数,构建支持向量机和神经网络模型对混合料养生性能进行预测,对两个模型的预测精度进行对比分析.结果表明:支持向量机模型预测误差率集中在2%以内,最大误差率为5.6%;神经网络模型的预测误差率集中在0~5%之间,最大误差率为19%,支持向量机模型的预测效果更为精准.In order to reveal the intrinsic relationship between the curing conditions and the curing properties of cold recycled mixed materials of emulsified asphalt,a prediction model for the curing properties of cold recycled specimens of emulsified asphalt was established with different algorithms.The curing conditions were taken as the model input parameters,and the support vector machine(SVM)and the neural network models were established to predict the curing properties of mixed materials.In addition,the prediction accuracy of two models was compared and analyzed.The results show that the prediction error rate of SVM model is concentrated within 2%,and the maximum error rate is 5.6%.The prediction error rate of neural network model is concentrated between 0 and 5%,and the maximum error rate is 19%.The prediction effect of SVM model is more accurate.

关 键 词：道路工程 沥青路面 乳化沥青 养生条件 养生性能 支持向量机 BP神经网络 预测精度 

分 类 号：TU528.42[建筑科学—建筑技术科学]