摘要
目的分析影响沥青混合料低温弯拉应变的因素,预测沥青混合料低温弯拉应变.方法基于Mablab7.1平台,应用灰色关联熵法分析了影响沥青混合料低温弯拉应变的因素,建立了结构为9-14-1的三层沥青混合料低温弯拉应变的BP神经网络预测模型.结果根据灰色关联熵法分析,确定了5℃延度、针入度指数PI、当量脆点t1.2、当量软化点t800、FAc比、25℃针入度、FAf比、软化点、CA比等9个影响因素作为神经网络模型的输入因素.通过43组试验数据对BP神经网络模型进行了学习训练,并用另外5组试验数据对模型进行了检验.预测结果与实测结果误差在工程要求精度范围以内.结论预测结果与实测结果的拟和程度较高,预测模型可信,可用于沥青混合料低温性能预测.
Limiting flexural strain at low temperature is the key parameter to reflect the low temperature crack resistance of asphalt mixtures. Precisely predicting the limiting flexural strain at low temperature is of great importance in the design of asphalt mixtures and production. With the aid of grey entropy grade analysis, the parameters that affect the limiting flexural strain at low temperature were analyzed. Prediction model of the limiting flexural strain at low temperature was set by BP neural network. The model was trained by 43 test data, and was proved by other 5 test data. Practical application demonstrated that the prediction model met the practical use.
出处
《沈阳建筑大学学报（自然科学版）》
CAS
北大核心
2009年第2期224-229,共6页
Journal of Shenyang Jianzhu University：Natural Science
基金
国家十一五科技支撑计划重点项目(2006BAJ18B05)
教育部博士点基金项目(20060213002)
关键词
沥青混合料
低温弯拉应变
BP神经网络
预测模型
灰色关联熵
asphalt mixture
flexural strain at low temperature
BP neural network
prediction model
grey entropy relation grade
