基于BP神经网络的硬化混凝土水胶比预测  

Prediction of hardened concrete water cement ratio based on BP neural network

作　　者：李悦[1,2] 郭伟[1,2] 杨进波 

机构地区：[1]北京工业大学城市与工程安全减灾教育部重点实验室,北京100124 [2]北京工业大学工程抗震与结构诊治北京市重点实验室,北京100124 [3]北京市顺义区住房和城乡建设委员会,北京101300

出　　处：《混凝土》2015年第8期1-4,共4页Concrete

基　　金：国家自然科学基金"硬化混凝土原始组分的分析原理与方法"(51278014);教育部"新世纪优秀人才支持计划"项目(NCET-12-0605);北京市属高等学校高层次人才引进与培养计划项目(CIT&TCD20150310)

摘　　要：混凝土一旦凝结硬化,再次评价其原始组分非常困难,目前针对水胶比的测试方法,试验复杂,精度不高。而水胶比是混凝土的重要参数,对混凝土的性能发挥至关重要的作用。试验制备了不同水胶比的混凝土,测试了不同龄期时混凝土强度、水泥浆钙元素含量、粗细集料用量。由于水胶比与上述因素间存在相关性,采用BP神经网络分析了硬化混凝土强度、龄期、粗细集料用量、水泥浆钙元素含量与硬化混凝土水胶比的非线性映射关系。研究结果表明:BP神经网络模型预测结果与实际水胶比吻合度较高,具有良好的预测效果。Once the concrete has hardened, the reevaluation of its original components will be very difficult.The current test methods for water/cement ratio is complex, but the accuracy is not high. The water cement ratio is an important parameter of concrete and it plays an important role on the performance of concrete. In the experiment, different water/cement ratio of concrete were prepared. Testing data in- cluding the concrete strength in different age, the content of calcium ion in cement slurry, the amount coarse aggregate and fine aggre- gate. As the water/cement ratio associated with these factors, BP neural network prediction method are adopted to establish the nonlinear mapping relationship between water/cement ratio and the concrete strength, the age and the aggregate consumption. The results show that the error predicted by BP neural network is small, and the model has a good predictive effect.

关 键 词：BP神经网络 硬化混凝土 水胶比 预测 

分 类 号：TU528.062[建筑科学—建筑技术科学]