Development of Hybrid Gravitational Search Algorithm Based Support Vector Regression for Predicting Magnetic Ordering Temperature of Manganite using Ionic Radii Descriptors
Abajingin, D.D. and *Owolabi, T.O.
Abstract
Unique characteristic features of Magnetic Refrigeration Technology (MRT) which include environmental
friendliness and better efficiency have promoted the chance of replacing the conventional Compression Gas System
of Refrigeration (CGSR). One of the main challenges associated with practical implementation of MRT is the tuning
of its magnetic ordering temperature (TC) to ambient value. The present work utilizes the ionic radii as potential
descriptors to develop hybrid Support Vector Regression (SVR) and Gravitational Search Algorithms (GSA) based
model for estimating magnetic ordering temperature. The outcomes of the proposed method agreed closely with the
experimental values. With the outstanding performance of the proposed hybrid SVR-GSA model, the magnetic
ordering temperature of manganite (which serves as the refrigerant) can be easily tuned to the ambient value while
dependency on ozone-depleting CGSR can be minimized.