基于支持向量回归的混凝土抗压强度预测研究  

Research on Prediction of Concrete Compressive Strength Based on Support Vector Regression

作　　者：唐江凌[1] TANG Jiang-ling(Guilin Normal College, Department of physics and Engineering Technology, Guilin 541199, China)

机构地区：[1]桂林师范高等专科学校物理与工程技术系

出　　处：《电脑知识与技术》2019年第6Z期268-269,272,共3页Computer Knowledge and Technology

基　　金：广西高校中青年教师基础能力提升项目“支持向量回归技术在混凝土强度预测中的应用研究”(项目编号:2018KY0916)

摘　　要：为了预测不同因素影响下的混凝土抗压强度,根据实测数据集,应用支持向量回归(SVR)方法,建立了SVR预测模型。模型以混凝土等效水灰比((mW/mC)E)、骨料平均浆体厚度(APT)、粉煤灰与胶凝材料用量比(βF)等3个参数为输入变量,以混凝土28d抗压强度为输出变量(fcu,28)进行学习和预测。经过与人工神经网络(ANN)预测模型进行比较,结果表明:对于相同的训练样本和检验样本,支持向量回归的预测模型比人工神经网络模型具有更高的预测精度。For prediction the concrete compressive strength at different factor, based on experimental data, use the support vector regression (SVR) approach to establish a SVR prediction model. For the model, equivalent water-cement ratio of concrete, average paste thickness of aggregates, fly ash-binder ratio act as input parameters and 28d concrete compressive strength were as output parameters. Compares with artificial neural network (ANN), it was found: the precision is higher for support vector regression model by applying identical samples.

关 键 词：支持向量回归 混凝土 等效水灰比 骨料平均浆体厚度 粉煤灰与胶凝材料用量 

分 类 号：TU528.01[建筑科学—建筑技术科学]