Application of generalized regression neural network in prediction of cement properties

Different modelling methods based on neural networks have become popular and been widely used in a great variety of fields. The properties of cements were predicted by neural network according to their chemical compositions, fineness, and other factors in this paper. The results showed that generalized regression neural network (GRNN) has higher accuracy and faster training speed compared with BP neural network. The maximum relative errors of the hydration heat and compressive strength predicted by GRNN were less than 5% and 9%, respectively.