Scholarlink

<p>A<sub>2</sub>B<sup>′</sup>BO<sub>6</sub> (where A is the rare-earth metal, B is a transition metal, and B<sup>′</sup> is any metal) double perovskite-based oxides exhibit novel and interesting features such as exchange bias, multiferroic and magnetocaloric effect, which promote their candidature for magnetic cooling applications. The double exchange interaction (existing between B′-O-B) that influences the observed magnetocaloric effect for cooling applications becomes strengthened by the proper combination of electronic configuration and ionic radii of rare-earth and B<sup>′</sup> and B ions. In this contribution, the magnetocaloric effect of the A<sub>2</sub>B′BO<sub>6</sub> system, a double perovskite, is modeled using random forest regression (RFR) and hybrid genetically optimized support vector regression (GSVR) algorithms, which employ crystal lattice parameters and ionic radii as predictors at various applied magnetic fields. Using assessment parameters such as correlation coefficient (CC), root mean square error (RMSE), and mean absolute error (MAE), the performances of the developed GSVR-ionic and RFR-ionic models with ionic radii descriptors are compared with those of the GSVR-latt and RFR-latt models, which employ crystal structural parameters as predictors. In the case of the RFR-based model, the values of assessment parameters computed for training samples of double perovskite magnetocaloric oxides using RFR-IONIC were 0.9873, 1.3750, and 1.1069 J/Kg&#xa0;K corresponding to CC, RMSE, and MAE, respectively. For the same assessment parameters using the RFR-latt model, 0.9096, 2.6041, and 1.7635 J/Kg&#xa0;K were, respectively, computed. The GSVR-ionic model outperforms the GSVR-latt, RFR-ionic, and RFR-latt models, achieving improvements of 8.18, 57.63, and 47.85%, respectively, as measured by the RMSE metric. Performance superiorities of 4.89, 60.13, and 40.95% were obtained using the MAE performance metric with double perovskite magnetocaloric oxide testing samples. The outstanding performance demonstrated by the models developed would strengthen the exploration of the magnetocaloric effect in the A<sub>2</sub>B′BO<sub>6</sub> system of double perovskite for addressing energy crises in cooling applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Modeling of Magnetocaloric Effect in A2B′BO6 Transition Metal Double Perovskite-Based Oxides for Cooling Applications Using Random Forest and Hybrid Support Vector Regression Intelligent Methods

  • 期刊论文

  • 发表日期：

    2025年8月19日

查看全文
  • Sami M. Ibn Shamsah

摘要

A2B′BO6 (where A is the rare-earth metal, B is a transition metal, and B′ is any metal) double perovskite-based oxides exhibit novel and interesting features such as exchange bias, multiferroic and magnetocaloric effect, which promote their candidature for magnetic cooling applications. The double exchange interaction (existing between B′-O-B) that influences the observed magnetocaloric effect for cooling applications becomes strengthened by the proper combination of electronic configuration and ionic radii of rare-earth and B′ and B ions. In this contribution, the magnetocaloric effect of the A2B′BO6 system, a double perovskite, is modeled using random forest regression (RFR) and hybrid genetically optimized support vector regression (GSVR) algorithms, which employ crystal lattice parameters and ionic radii as predictors at various applied magnetic fields. Using assessment parameters such as correlation coefficient (CC), root mean square error (RMSE), and mean absolute error (MAE), the performances of the developed GSVR-ionic and RFR-ionic models with ionic radii descriptors are compared with those of the GSVR-latt and RFR-latt models, which employ crystal structural parameters as predictors. In the case of the RFR-based model, the values of assessment parameters computed for training samples of double perovskite magnetocaloric oxides using RFR-IONIC were 0.9873, 1.3750, and 1.1069 J/Kg K corresponding to CC, RMSE, and MAE, respectively. For the same assessment parameters using the RFR-latt model, 0.9096, 2.6041, and 1.7635 J/Kg K were, respectively, computed. The GSVR-ionic model outperforms the GSVR-latt, RFR-ionic, and RFR-latt models, achieving improvements of 8.18, 57.63, and 47.85%, respectively, as measured by the RMSE metric. Performance superiorities of 4.89, 60.13, and 40.95% were obtained using the MAE performance metric with double perovskite magnetocaloric oxide testing samples. The outstanding performance demonstrated by the models developed would strengthen the exploration of the magnetocaloric effect in the A2B′BO6 system of double perovskite for addressing energy crises in cooling applications.