Abstract
Magnetic refrigeration (MR) combines many unique features such as low cost, high efficiency and environmental friendliness which make it preferred to the conventional gas compression system of refrigeration. MR employs manganite-based material due to its high magnetocaloric effect as well as tunable Curie temperature (\({T}_{\mathrm{C}}\)). For effective utilization of this technology, \({T}_{\mathrm{C}}\) of manganite refrigerant needs to be tuned to ambient room temperature. In order to relieve experimental stress involved and consequently save valuable time and resources, support vector regression (SVR) computational intelligence technique is proposed using manual search (MS-SVR) and a novel gravitational search algorithm (GSA-SVR) for its hyper-parameter optimization. The developed GSA-SVR model shows better performance than MS-SVR model with performance improvement of 86.03% on the basis of root mean square error (RMSE) and 0.07% on the basis of correlation coefficient (CC) on the training dataset while 11.48% of RMSE improvement and 2.48% of CC improvement were recorded for the testing dataset. The outstanding results presented in this work suggest the potential of the proposed models in promoting room temperature MR through quick estimation of the effect of dopants on \({T}_{\mathrm{C}}\) so as to obtain manganite that works well around the room temperature without loss of precision.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have