基于BP神经网络的陶瓷裂纹釉釉面效果预测模型及应用  

BASED ON BP NEURAL NETWORK POTTERY CRACKLE GLAZE’S FORECAST MODELS AND APPLICATION

作　　者：柳炳祥[1] 洪晶[1] 方泽军[2] 程功勋[1] 

机构地区：[1]景德镇陶瓷学院,景德镇333000 [2]北京工业大学,北京100022

出　　处：《中国陶瓷》2007年第4期31-34,共4页China Ceramics

摘　　要：人工神经网络具有巨量并行、结构可变和高度非线性等特点,其建立数学模型并不需要预先知道太多有关问题背景的知识,这尤其适用于陶瓷釉研究中某些机理尚未完全清楚、传统数学方法无法分析的情况[1]。将人工神经网络技术用于裂纹釉的配方性能分析,以釉面裂纹为研究对象,选取了8种釉料的化学成分,在均匀实验设计的基础上,用BP人工神经网络对所得实验结果进行了分析,并且用图形化方式直观地表达了出来。根据实验结果,人工神经网络模型能较准确地预测出陶瓷的釉面效果,从而为研究裂纹釉提供了一种新的思路和有效手段。Artificial neural network with massive parallel, structure variable and highly nonlinear characteristics of the establishment of mathematical models that did not require too much advance knowledge of the background issues, this research was particularly applicable to ceramic glaze certain mechanisms had not yet entirely clear, traditional mathematical methods to analyze the situation. This paper applied Artificial Neural Network technology to analyze crackle glaze recipes performance and study crackle glaze, selected the eight Egyptian chemical composition, in homogeneous experimental design basis, using BP Neural Network analysis of the results obtained from experiments, and used a graphic way to express them intuitively. Based on laboratory test results, ANN model can predict the effects of glazed pottery more accurately, and provided a new way of thinking and effective means for crackle glaze＇ s research.

关 键 词：裂纹釉 人工神经网络 BP神经网络 釉面效果 裂纹 预测 

分 类 号：TQ174.43[化学工程—陶瓷工业]