Authors : Abdullah Alqahtani, Taoreed O. Owolabi, Kabiru O. Akanded, Sunday O. Olatunji and Nahier Aldhafferi
Abstract: Manganite based materials have several unique properties which include low cost, environmental friendliness and huge magnetocaloric effect which make them suitable as magnetic refrigerant in Magnetic Refrigeration Ttechnology (MRT). However, effective utilization and deployment of this technology requires manganite refrigerants with magnetic ordering Temperature (T ) around ambient temperature. In order to tune C the T of manganite based materials to the desired ambient value and circumvent experimental stress therein, C a hybrid of Support Vector Regression (SVR) Algorithm and Genetic Algorithm (GA) is proposed using the crystal lattice structural parameter of manganite based materials as the descriptors to the models. The generalization and predictive strength of the proposed hybrid GA-SVR Model is compared with the existing Gravitational Search Algorithm based model (GSA-SVR) on the basis of error reduction as well as computational complexity. The proposed hybrid GA-SVR Model outperforms the existing GSA-SVR Model with 8% and 2% performance improvement on the basis of the coefficient of correlation and root mean square error, respectively. Comparison of the proposed hybrid GA-SVR Model with the existing Manual Search based Support Vector Regression (MS-SVR) shows that the developed hybrid GA-SVR model outperforms MS-SVR Model with percentage improvement of 4.7% on the basis of correlation coefficient. Better performance demonstrated by GA-SVR Model coupled with its less computational time, strengthens its potential in enhancing room temperature magnetic refrigeration and promoting environmental friendly technology over ozone-depleting refrigeration technology.
Abdullah Alqahtani, Taoreed O. Owolabi, Kabiru O. Akanded, Sunday O. Olatunji and Nahier Aldhafferi, 2018. Hybrid Support Vector Regression and Genetic Algorithm Model for Tuning Magnetic Ordering Temperature of Manganite Refrigerant. Research Journal of Applied Sciences, 13: 87-93.