基于ANN的TA15钛合金TIG焊接头拉伸性能模型建立

ANN prediction models for tensile properties of TIG welded joints of TA15 titanium alloys

摘要: 近α型钛合金TA15具有较好的综合力学性能,在飞机结构中有广阔的应用前景。针对TA15钛合金进行了系统的TIG焊试验,并进行了焊接接头拉伸性能测试。在此基础上,基于人工神经网络(ANN)原理对影响TA15钛合金TIG焊接头拉伸性能的主要参数进行了选取,确定ANN的输入和输出参数,建立了TA15钛合金TIG焊接头拉伸性能ANN模型并对模型进行了优化及误差分析。结果表明,最终建立的拉伸性能模型能很好地适用于TA15钛合金TIG焊接头拉伸性能的模拟,优于传统回归方法。

Abstract: Near-alpha TA15 titanium alloy has good mechanical properties which has a wide application in airframe.Sy stemic TIG (tungsten inert-gas)welding experiments were performed for TA15 titanium alloy and the tensile tests of their welded joints were also carried out.Main factors were then selected after relating the parameters to tensile properties based on the principle of artificial neural network (ANN).Finally, the input and output parameters were confirmed and the ANN prediction models for tensile properties of titanium alloys TA15 TIG welded joints were established.Those models were then optimized and the model errors were analyzed.It showed that the prediction models, better than the traditional regression method, could be well used to predict tensile properties of titanium alloys TA15 TIG welded joints.