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。 In order to correctly identify ancient and modern name porcelain, 40 samples of Jun porcelain fragments were selected from ancient Kwan Tai kiln and modern Kongjia Jun kilns, Mijia jun kilns and Xinghang Jun kilns, and each specimen was measured by INAA , The contents of 25 elements in the glaze, the INAA data of these samples were trained and predicted by BP neural network. The prediction results show that the well-trained network can better identify the fetal and enamel samples of unknown ancient Jun porcelain and modern high-grade Jun porcelain.