Modeling magnetocaloric effect of doped EuTiO3 perovskite for cooling technology using swarm intelligent based support vector regression computational method

Abstract

This work develops particle swarm optimization based support vector regression (PSVR) algorithm for predicting maximum magnetic entropy change using ionic radii of the constituting elements as well as the applied magnetic field as descriptors. The developed PSVR-G using Gaussian kernel function performs better than PSVR-P which employs polynomial data-mapping function with performance improvement of 0.04%, 9.93% and 5.25% on the basis of correlation coefficient (CC), mean absolute error (MAE) and root mean square error (RMSE), respectively on testing data samples. The influence of magnetic field ranging between 1 T and 5 T on doped perovskite EuTiO3 was investigated for different samples using the developed PSVR-G model. The significance of various dopants on perovskite EuTiO3 was further examined using the developed PSVR-G model. The accuracy and precision associated with the developed models would facilitate comprehensive exploration of perovskite EuTiO3 material for magnetic cooling application without experimental stress.