Prediction of Ceramic’s Mechanical Properties Based on Sintering Temperature using Neural Network

Abstract

Ceramics is one of material which apply in many area.  Thus, study of its properties is very important to fulfilled the properties requirement. The mechanical properties of ceramic such as flexural strength and hardness mainly depend on the sintering temperature and additive material. The experiments must be done to determine the best mechanical properties based on proportional sintering temperature and additive materials. Simulation for predicting mechanical properties of ceramics had been developed by using Artificial Neural Network. According to neural network simulation, the graphic of simulation result had same pattern to experimental data as the target. For predicting hardness, the Normalized Root Mean Square Error of network is 0 at training and 0.077 at validation part. This value is in line to its Coefficient Correlation which have value closed to 1. Meanwhile, the network can be used to predict flexural strength of ceramics excellently.