基于BP神经网络的粘结NdFeB永磁体性能预测  

Properties prediction of Bonded NdFeB Permanent Magnet on BP Neural Network

作　　者：储林华[1] 查五生[1] 刘锦云[1] 陈德波[1] 

机构地区：[1]西华大学材料科学与工程学院,四川成都610039

出　　处：《材料科学与工程学报》2008年第6期958-962,共5页Journal of Materials Science and Engineering

基　　金：四川省教育厅重点资助项目(2004A110)

摘　　要：本文基于MATLAB平台和现有的少量实验数据,利用神经网络的非线性映射和泛化能力,建立了一个输入为工艺参数、输出为NdFeB永磁体性能参量的BP神经网络模型,并通过检验样本检验了ANN模型的准确性。实验表明:将神经网络技术应用于材料制备工艺设计,可以明显缩短实验周期,提高工艺设计效率,对实际的研究工作具有一定的指导意义和应用价值。According to the MATLAB terrace and the limited amount of experimental data, a BP neural network model of the bonded NdFeB magnet is built on neural network for its nonlinear mapping and generalization abilities. The model takes the craft parameter as input, and the magnetic property parameters as output. The accuracy of ANN model is tested by the test sample. The experimental results show that the way using ANN network technology in material preparation crafts optimizing, can obviously shorten the experimental period and improve the craft designing efficiency. Therefore, the modeling method is effective and the model is available.

关 键 词：BP神经网络 NDFEB粘结磁体 磁性能 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]