Scholarlink

<p>Thermomechanical analysis (TMA) was used to collect a large dataset on ancient ceramic sherds, which was analyzed using hierarchical cluster analysis (HCA) and k-means clustering to investigate the suitability of this technique for the convenient and automatic comparison of large numbers of samples, since manual analysis and comparison of TMA curves are time-consuming procedures; the study found that the best results for classifying homogeneous sample groups were obtained by using the coefficient of thermal expansion (CTE), shrinkage, and firing conditions as inputs for the clustering algorithms. Various material characteristics including softening, melted, phase transformation, and solid-state reactions can be studied in this way. For this reason, several types of ceramic sherds were examined, representing earthenware, porcelain, and stoneware ceramic from various regions and of various ages. Along with being a very helpful technique in reducing the dimensionality of the dataset, this also allowed us to assign one sample in each set as the most representative sample. Furthermore, a cluster analysis of TMA data is an effective way to investigate the firing dynamics in terms of deformation temperature and to determine when shrinkage happens, according to a statistical comparison of thermomechanical analysis trials. In addition to these benefits, the shortcomings and limitations of this statistical technique were also evaluated in the TMA data obtained from ancient pottery sherds.</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

Clustering and Thermal Analysis of Ancient Ceramic Artifacts through TMA Data

  • 期刊论文

  • 发表日期：

    2026年7月23日

查看全文
  • Md Saifur Rahman,
  • Hendrik Simon Cornelis Metselaar,
  • Bushroa Binti Abdul Razak

摘要

Thermomechanical analysis (TMA) was used to collect a large dataset on ancient ceramic sherds, which was analyzed using hierarchical cluster analysis (HCA) and k-means clustering to investigate the suitability of this technique for the convenient and automatic comparison of large numbers of samples, since manual analysis and comparison of TMA curves are time-consuming procedures; the study found that the best results for classifying homogeneous sample groups were obtained by using the coefficient of thermal expansion (CTE), shrinkage, and firing conditions as inputs for the clustering algorithms. Various material characteristics including softening, melted, phase transformation, and solid-state reactions can be studied in this way. For this reason, several types of ceramic sherds were examined, representing earthenware, porcelain, and stoneware ceramic from various regions and of various ages. Along with being a very helpful technique in reducing the dimensionality of the dataset, this also allowed us to assign one sample in each set as the most representative sample. Furthermore, a cluster analysis of TMA data is an effective way to investigate the firing dynamics in terms of deformation temperature and to determine when shrinkage happens, according to a statistical comparison of thermomechanical analysis trials. In addition to these benefits, the shortcomings and limitations of this statistical technique were also evaluated in the TMA data obtained from ancient pottery sherds.