Investigation of BaTiO3 formulation: an artificial neural network (ANN) method

Artificial neural networks (ANNs) are relatively new computational tools and their inherent ability to learn and recognize highly non-linear and complex relationships makes them ideally suited in solving a wide range of complex real-world problems. However, very few is known of the use of this technique in ceramics although it is often invoked in diverse areas in chemistry. Here application of ANN technique to model the BaTiO3 based dielectric ceramic formulation was carried through. Based on the homogenous experimental design the experimental results of 21 samples were analyzed by a three-layer back propagation (BP) network. Through comparison we found that the ANN model is much more accurate than conventional multiple nonlinear regression analysis (MNLR) model for the same set of data. The results of ANN model were also expressed and analyzed by intuitive graphics. It indicates that the three-layer BP network based modeling is a very useful tool in dealing with problems with serious non-linearity encountered in the formulation design of dielectric ceramics.