Scholarlink

<p>The paper presents models for a&#xa0;prediction of the Vickers microhardness in the indentation load range of 100 to 500 gf. Machine learning methods (random decision forest and adaptive boosting) are used to develop these models for ceramic materials based on hydroxyapatite with the addition of reinforcing multi-walled carbon nanotubes in the amount of&#xa0;0.1 and 0.5 wt.%. The input data for the model development include the sample type, test load and indenter diameter. A&#xa0;comparative analysis of training quality metrics such as the mean absolute, root mean square, and mean absolute percentage errors and the determination coefficient help to identify the method providing the lowest prediction error. The results of the regression analysis show that the adaptive boosting model with the high (0.918) determination coefficient and low (10.135) mean absolute error demonstrate the highest training accuracy. The proposed predictive models allow to determine the best machine learning method for predicting the Vickers microhardness of ceramic materials based on hydroxyapatite with regard the obtained nonlinear dependencies.</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

Microhardness modeling of hydroxyapatite ceramics using random forest methods

  • 期刊论文

  • 发表日期：

    2025年11月1日

查看全文
  • A. E. Rezvanova,
  • A. N. Ponomarev,
  • B. S. Kudryashov,
  • M. I. Kochergin,
  • V. Yu. Pogudin

摘要

The paper presents models for a prediction of the Vickers microhardness in the indentation load range of 100 to 500 gf. Machine learning methods (random decision forest and adaptive boosting) are used to develop these models for ceramic materials based on hydroxyapatite with the addition of reinforcing multi-walled carbon nanotubes in the amount of 0.1 and 0.5 wt.%. The input data for the model development include the sample type, test load and indenter diameter. A comparative analysis of training quality metrics such as the mean absolute, root mean square, and mean absolute percentage errors and the determination coefficient help to identify the method providing the lowest prediction error. The results of the regression analysis show that the adaptive boosting model with the high (0.918) determination coefficient and low (10.135) mean absolute error demonstrate the highest training accuracy. The proposed predictive models allow to determine the best machine learning method for predicting the Vickers microhardness of ceramic materials based on hydroxyapatite with regard the obtained nonlinear dependencies.