基于BP神经网络的铝合金锻造性能预测  

Prediction of Forging Performance of Aluminum Alloy Based on BP Neural Network

作　　者：魏萌[1] 肖凌 WEI Meng;XIAO Ling(Changjiang Polytechnic,Wuhan 430074,China;School of Computer Science&Technology,Institute of Information Security,Huazhong University of Science and Technology,Wuhan 430074,China)

机构地区：[1]长江职业学院,湖北武汉430074 [2]华中科技大学计算机科学与技术学院信息安全研究所,湖北武汉430074

出　　处：《热加工工艺》2021年第21期105-107,112,共4页Hot Working Technology

基　　金：2018年度湖北省教育厅科学研究计划指导性项目(B2018564)。

摘　　要：利用BP神经网络建立了汽车用铝合金锻造性能的网络模型,并以24组试验数据为训练样本,以剩余6组作为验证样本,对模型进行了训练和预测结果验证。结果表明,该模型输出的抗拉强度和磨损体积的相对平均训练误差分别为3.89%和4.04%,相对平均预测误差分别为3.86%和3.67%。该模型预测误差小,预测精度高。The network model of forging properties of the aluminum alloy for automobile was established by using BP neural network.Taking 24 groups of test data as training samples and the remaining six groups as validation samples,the model was trained and predicted result verificated.The results show that the average relative training errors of tensile strength and wear volume of the output of the model are 3.89%and 4.04%respectively,and the average relative prediction errors are 3.86%and 3.67%respectively.The prediction error of the model is small and the prediction accuracy is high.

关 键 词：BP神经网络 预测 汽车用铝合金 锻造性能 

分 类 号：TG319[金属学及工艺—金属压力加工] TG146.21[一般工业技术—材料科学与工程]