ANN技术在复合材料胶接修理分析中的应用  

Application of Artificial Neutral Network to Bonding Repair Analysis of Composite Structure

作　　者：程起有[1] 朱银垂[1] 罗国仁[1] 韩振忠[1] 

机构地区：[1]中国直升机设计研究所,江西景德镇333001

出　　处：《直升机技术》2011年第2期20-24,共5页Helicopter Technique

摘　　要：基于改进的BP人工神经网络(ANN)建立了复合材料胶接修理分析模型,结合采用复合材料胶接修理的正交试验及有限元分析的结果为训练和检测神经网络提供样本,有效地利用了神经网络、试验设计技术与有限元分析的优点。胶接修理实例分析结果表明,所建神经网络模型对胶接参数与修理效果之间关系的预测与试验结果一致,说明将人工神经网络应用于复合材料胶接修理参数分析是一种行之有效的方法。To assist the parameter selection in bonding repair of composite laminates,an analytic model based on back-propagation Artificial Neural Network(ANN) was established in this paper.The data of experiment and the finite element were used to get the training and testing set for Back-Propagation neural network prediction model.The advantages of artificial neural network,experimental design technique and finite element method were exploited.The Back-Propagation neural network was used to set up a prediction model for predicting bonding parameter.A good agreement between the prediction value and the experimental data was achieved so the prediction model was effective and helpful for parameter selection in bonding repair of composite laminates.

关 键 词：复合材料 胶接修理 人工神经网络 有限元 

分 类 号：TB3[一般工业技术—材料科学与工程]