基于BP神经网络的绿色混凝土抗压强度预测模型  

Prediction model of green concrete compressive strength based on BP neural network

作　　者：徐国强[1,2] 苏幼坡[1,2] 韩佃利[1,2] 张静[1,2] 

机构地区：[1]河北联合大学建筑工程学院,河北唐山063009 [2]河北省地震工程研究中心,河北唐山063009

出　　处：《混凝土》2013年第2期33-35,49,共4页Concrete

基　　金：国家自然科学基金项目(51278164);唐山市科技计划项目(12140208A-1)

摘　　要：与普通混凝土相比,绿色混凝土具有成分复杂的特点,为了在多因素作用下更为准确地预测绿色混凝土的抗压强度,在分析三层BP神经网络原理的基础上,选择影响绿色混凝土抗压强度的7个指标,以66个抗压强度试验为示例,建立了三层BP神经网络抗压强度预测模型。验证样本的训练结果表明,该模型能够较准确地快速预测绿色混凝土的抗压强度,并通过对各指标的权重计算,确定了影响绿色混凝土抗压强度的主要因素。Compared with ordinary concrete, green concrete was characterized by complicated composition.In order to more accurately predict the green concrete compressive strength under the influence of multiple factors, seven indexes that affect the compressive strength of green concrete were selected based on the analysis of three-layer BP neural network principle.With 66 compressive strength tests as an example, a prediction model of compressive strength on three-layer BP neural network was established.The training results of verification samples showed that the model could quickly and accurately predict the compressive strength of the green concrete.Through calculating the weight of various index, the main factors that affect green concrete compressive strength were determined.

关 键 词：BP神经网络 绿色混凝土 抗压强度 预测模型 

分 类 号：TU528.01[建筑科学—建筑技术科学]