Scholarlink

<p>The resonance frequency of the Fe/MgO system has been modeled by investigating the influence of processing parameters, including MgO nanoparticle size (10–1000&#xa0;nm), pressing pressure (600–1250&#xa0;MPa), sintering temperature (0–900&#xa0;°C), sintering time (0–60&#xa0;min), and annealing atmosphere (air, nitrogen, or no annealing). The resonance frequency (0.9–3300&#xa0;kHz) was successfully predicted using machine learning (Extreme Gradient Boosting (XGBoost), Gradient Boosting (GB), and random forest (RF)) models. The models achieved an R-squared (R²) of 0.99, thereby explaining 99% of the variance. The predictive accuracy of the models was further assessed using Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE). High R² values and consistently low RMSE and MAE scores demonstrate the robustness of the models for modeling resonance frequency. This work highlights key factors for optimizing the resonance frequency of Fe/MgO systems. SHAP values and feature importance analyses identified time and temperature as the most influential parameters across all models. The notable impacts of time and temperature indicate that optimizing the sintering process can lead to significant enhancements in material performance. The critical role of sintering parameters in optimizing the resonance properties of Fe/MgO systems, paving the way for improved material performance in high-frequency magnetic 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

A machine learning approach to predicting resonance frequency in soft magnetic composites

  • 期刊论文

  • 发表日期：

    2025年12月4日

查看全文
  • Tarik Sadat

摘要

The resonance frequency of the Fe/MgO system has been modeled by investigating the influence of processing parameters, including MgO nanoparticle size (10–1000 nm), pressing pressure (600–1250 MPa), sintering temperature (0–900 °C), sintering time (0–60 min), and annealing atmosphere (air, nitrogen, or no annealing). The resonance frequency (0.9–3300 kHz) was successfully predicted using machine learning (Extreme Gradient Boosting (XGBoost), Gradient Boosting (GB), and random forest (RF)) models. The models achieved an R-squared (R²) of 0.99, thereby explaining 99% of the variance. The predictive accuracy of the models was further assessed using Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE). High R² values and consistently low RMSE and MAE scores demonstrate the robustness of the models for modeling resonance frequency. This work highlights key factors for optimizing the resonance frequency of Fe/MgO systems. SHAP values and feature importance analyses identified time and temperature as the most influential parameters across all models. The notable impacts of time and temperature indicate that optimizing the sintering process can lead to significant enhancements in material performance. The critical role of sintering parameters in optimizing the resonance properties of Fe/MgO systems, paving the way for improved material performance in high-frequency magnetic applications.