Scholarlink

Abstract <p>The study is devoted to the development of models for predicting the refractive index of ceramic composite materials based on hydroxyapatite with the 0.1 and 0.5 wt % multi-walled carbon nanotubes additives by machine learning (ML) methods. The refraction prediction is based on experimental study results of the porous structure of ceramics in the frequency range from 0.2 to 1.6 THz. A new methodology which includes the construction of models based on methods such as linear and polynomial approximation, random decision forest and artificial neural networks, has been developed to quantitatively assess the influence of carbon nanotubes on the refractive index of the composite. The results of application of the neural networks showed significantly higher forecasting accuracy, the average absolute error of which is ~0.04%. Our findings underscore the effectiveness of using machine learning in non-invasive analysis of the porous structure of heterogeneous composites.</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

Machine Learning for Analysis of a Porous Structure of Composite Ceramics Based on Hydroxyapatite with the Multi-Walled Carbon Nanotubes Additives

  • 期刊论文

  • 发表日期：

    2025年10月29日

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

摘要

Abstract

The study is devoted to the development of models for predicting the refractive index of ceramic composite materials based on hydroxyapatite with the 0.1 and 0.5 wt % multi-walled carbon nanotubes additives by machine learning (ML) methods. The refraction prediction is based on experimental study results of the porous structure of ceramics in the frequency range from 0.2 to 1.6 THz. A new methodology which includes the construction of models based on methods such as linear and polynomial approximation, random decision forest and artificial neural networks, has been developed to quantitatively assess the influence of carbon nanotubes on the refractive index of the composite. The results of application of the neural networks showed significantly higher forecasting accuracy, the average absolute error of which is ~0.04%. Our findings underscore the effectiveness of using machine learning in non-invasive analysis of the porous structure of heterogeneous composites.