BP神经网络对稀土元素掺杂钛酸钡基陶瓷的研究

Investigation Into Rare-earth Elements Doping in BaTiO_3 Ceramics Using a BP Artificial Neural Network

将Gd、Ce、Nd、Sm稀土氧化物的加入量作为输入,钛酸钡陶瓷材料介电常数及介电损耗作为输出,建立了BP神经网络模型,网络以traingdx函数作为学习训练函数,选取部分实验数据作为学习样本对网络进行了训练,最终得到了满意的性能预测值。

In this paper,a BP artificial neural network is developed with the content of rare-earth metal oxide dopants in BaTiO3 ceramics, such as Gd, Ce, Nd and Sm oxides, used as an input, and the dielectric constant and dielectric loss of the BaTiO3 ceramics as an output, respectively. A function of traingdx is applied for training, and some experimental data are employed as training samples for the network. The BP artificial neural network consequently perfectly predicts the dielectric property of BaTiO3 ceramics.