Prediction of relationships between hydrostatic coefficients and processing parameters of porous PZT ceramics by radial basis function

The high hydrostatic coefficients of porous piezoelectric materials provide the reduced acoustic impedance for the effective coupling with water medium as well as biological tissues. The hydrostatic strain coefficient (dh), voltage coefficients (gh) and longitudinal piezoelectric coefficient (d33) of porous piezoelectric materials has a strong dependence on processing parameters such as porosity parameter, polymer percentage, density and poling voltage. In this work, a nonparametric regression has been carried out to figure out the nonlinear relationships between the hydrostatic coefficients and its processing parameters. The nonparametric regression has been carried out by artificial neural network. The Network architecture utilized is radial basis function (RBF) which is combined with Fuzzy Clustering Method (FCM) to identify the hidden node numbers and widths. The porosity parameter was calculated from scanning electron micrographs of the porous piezoelectric materials using conventional image processing techniques. The results of the artificial neural network were evaluated against experimental data.