高强混凝土强度的人工神经网络预测

Forecast strength of high strength concrete with neural networks

为了在多因素作用下更为准确地预测高强混凝土的抗压强度,采用了人工神经网络中的BP网络模型及其学习算法,基于MATLAB神经网络工具箱对文献中高强混凝土的实测数据进行分析预测,并与回归分析方法计算的结果进行了对比,结果表明,人工神经网络在高强混凝土抗压强度的预测方面具有较高的精度,且明显优于回归模型.因此,神经网络方法是一种可以定量分析、简便易行、计算精度高、预测能力强的分析方法,用于高强混凝土抗压强度的预测是可行的.

To make the strength forecast of high strength concrete under influence of several factors exact, the model of BP network and its learning algorithms are recommended. Then the approach based on Matlab-NNT is applied to predict the strength of high strength concrete. Furthermore, we contrast it to the regression. It is found that BP network can predict the strength of high strength concrete more accurately than the approach of regression does. The result suggests that neutral network is a quantitative and convenient analyzing approach with high accuracy. It is feasible in predicting the strength of high strength concrete.