论文部分内容阅读
用EDXRF探针测定了48个来自3个不同产地的4个古代青瓷窑址的瓷片的化学组成.接着用SOM神经网络对胎的主、次量元素化学组成数据进行了聚类分析,结果表明SOM能正确地区分出瓷片的3个产地:龙泉、慈溪和杭州.由于两官窑的瓷片在主、次量元素化学组成上比较接近,使得它们的分类正确率仅为76.92%.于是,作者对两官窑瓷片胎的微量元素化学组成数据用同样的方法又进行了一次聚类分析,结果发现分类正确率提高到了84.61%,说明两官窑的瓷片在微量元素上的差别比在主、次量元素上的差别要大,这与实际情况是一致的.研究表明SOM神经网络方法可以应用于古陶瓷的聚类分析研究中。
The chemical composition of 48 porcelain tiles from 48 ancient celadon kiln sites in 3 different places was measured by EDXRF probe.And then SOM neural network was used to cluster the primary and secondary chemical composition data of the fetus and the results This shows that SOM can correctly distinguish the three producing areas of porcelain tiles: Longquan, Cixi and Hangzhou.Because the chemical composition of the two kilns is close to that of the primary and secondary elements, their classification accuracy is only 76.92%. Therefore, the author of the two official kiln ceramic tile trace element chemical composition data using the same method and a cluster analysis, the results showed that the classification accuracy increased to 84.61%, indicating two kiln ceramic tiles in the trace elements The difference is greater than the difference between primary and secondary elements, which is consistent with the actual situation.Research shows that the SOM neural network method can be applied to the cluster analysis of ancient ceramics.