Prediction of effects of microstructural phases using generalized regression neural network

In the scope of this study, microstructure–macroproperty relationship of cement mortars has been established in order to define the effects of microstructural phases on strength. Microstructural studies have been become great issue in materials engineering. Nowadays, to characterize the microstructural phase properties and to improve and modify them are performed by scientist to forecasting and enhancing. According to this objective, cement mortars incorporating with chemical admixtures were prepared to constitute different microstructural graphs. These micrographs were analyzed to determine the amounts of unhydrated cement part, undifferentiated hydrated part and capillary pore phases in the cement mortar sections. Afterwards, the amounts of these microstructural phases were related to strength values of each cement mortar specimen. The relationship was established by using generalized regression neural network analysis.

▸ Microstructure–macroproperty relationship is important to modify macroproperties. ▸ GRNN analysis is one of the popular methods to establish this relationship. ▸ This study indicates the availability and effective capacity of GRNN on this relationship.