基于神经网络BP算法的7075-T651铝合金搅拌摩擦焊焊接接头疲劳寿命预测  

Fatigue life prediction of friction-stir welding joints of aluminum alloy 7070-7651 based on BP algorithm of neural network

作　　者：王希靖[1] 徐成[1] 张杰[1] 李树伟[1] 牛勇[1] 

机构地区：[1]兰州理工大学甘肃省有色金属新材料重点实验室,甘肃兰州730050

出　　处：《兰州理工大学学报》2008年第3期12-15,共4页Journal of Lanzhou University of Technology

基　　金：国家自然科学基金(10577010)

摘　　要：在对航空用5 mm厚7075-T651铝合金搅拌摩擦焊焊接接头进行低周疲劳实验研究的过程中,通过实验数据的获取,建立基于人工神经网络BP算法的搅拌摩擦焊焊接接头疲劳寿命预测模型.利用该网络模型对不同工艺参数下焊接接头的疲劳寿命进行预测.预测值和实验值比较表明,预测数据与实验数据吻合良好.该模型的建立为铝合金搅拌摩擦焊焊接接头疲劳寿命研究节省了时间.By means of experimentally investigation of low-cycle fatigue of friction-stir welded joints of aviation aluminum alloy 7075-7651 with thickness of 5 mm and according to the experiment data obtained, a prediction model of fatigue life of the joints was built up on the basis of the BP algorithm of artificial neural network. By using it, the fatigue life of the welding joints with different processing parameters was predicted. It was shown by the comparison of the predicted data to those of experiments that they were well consistent with each other, so that the time consumed on the investigation of the fatigue life of such joints was saved by using this model.

关 键 词：搅拌摩擦焊 疲劳寿命 神经网络BP算法 学习样本 

分 类 号：TG407[金属学及工艺—焊接]