结合多元回归的BP网络对HPC强度的预估

Predicting the HPC Strength Using BP Neural Network Combining with Multi-member Regression

对于高性能混凝土(HPC)强度的预估,传统的多元线性拟合的方法存在模型适用度差,预估精度不高等缺点.笔者将BP网络技术应用在高性能混凝土强度的预估中,并结合多元线性回归,采用模拟样本的方法建立了强度预估的BP网络模型.预估结果表明,该方法是非常有效可靠的,且具有简单、适应性好等特点.

For the prediction of the HPC strength, the method of multi-member linearity regression has some defects that the applicability of models is bad and the prediction precision is lowly. The means of back-propagation neural network was used in this paper to do the same work. Combining with the multi-member linearity simulation,a BP network model base on the method of simulative sample was founded. The prediction results show that the method is effective and dependable as well as good adaptability.