Component Analysis of Civil Building Materials Using Artificial Neural Networks

摘要

This paper is concerned with the use of artificial neural networks (ANN) aimed at component analysis of civil building materials. Based on radial basis function neural network (RBFN) and perception neural network, a four-layer feed-forward neural network named radial basis perception network (RBPN) is built. The input information is mapped into output through two hidden layers where the units are connected when necessary. The connected criterions are decided by the information included in the output samples. CaO-Al_2O_3- SiO_2 system is discussed in this paper as an example. During the learning procedure, the input and output information are all used to realize model cluster and an adjustable parameter σ of RBF centers is available. Simulation shows that neural network can be used to predict the compound components of civil building materials successfully and this method is convenient and gets good generalization ability.