Estimation of critical current density of bulk superconductor with artificial neural network

In the applications of superconducting materials, the critical current density $J_c\left(B\right)$ is a crucial performance parameter. The conventional method of measuring $J_c\left(B\right)$ of bulk superconductor is magnetization method. However, there are errors in the estimation of $J_c\left(B\right)$ in the lower field, and the estimation is not applicable in the region where the magnetic field reverses. In this paper, $J_c\left(B\right)$ of the bulk superconductor is determined by the hysteresis and magnetostriction loops with artificial neural network (ANN), respectively. Compared with double-output ANN, the critical current density obtained by single-output ANN is more accurate. Finally, the prediction results given by the hysteresis and magnetostriction loops are discussed.