古今名钧瓷鉴别的INAA和BP神经网络研究

Distinguish of Famous Jun Porcelain in Ancient and Present Age by INAA and BP Neural Network

摘要: 为正确鉴别古今名钧瓷，从古钧台窑和现代孔家钧窑、苗家钧窑、星航钧窑选取钧瓷残片样品40个，用仪器中子活化分析（INAA）测得每个样品胎、釉中25种元素的含量，应用BP神经网络对这些样品的INAA数据进行了训练和预测。预测结果表明，训练好的网络能较好鉴别未知古钧官瓷和现代高档钧瓷的胎、釉样本。

Abstract: Forty samples of Jun porcelain from an ancient Juntai kiln and 3 modern Jun kilns (Kongjia, Miaojia and Xinghang) were selected and analyzed for 25 elements by INAA. The data were trained and forecasted by BP neural network. The results indicate that the network can distinguish unknown body and glaze samples of the official Jun porcelain and the modern top-grade Jun porcelain after proper training.