基于人工神经网络预测铝基复合材料瞬间液相扩散焊的接头性能

Forecasting of Joint Performance of TLP Diffusion Bonding for Aluminium-based Composite Using Artificial Neutral Network

非连续增强铝基复合材料的瞬间液相(TLP)扩散焊的焊接工艺与其接头力学性能之间具有很强的非线性关系,人工神经网络是解决非线性映射关系的一种有效手段。本文以Al2O3p/6061Al的TLP扩散焊(用Cu箔作中间层)焊接工艺与接头抗剪强度的关系为研究对象,在Matlab语言环境下,以正交实验数据作为训练和预测样本,用5节点的单隐含层BP型神经网络进行了预测。结果表明:正交实验和人工神经网络相结合来预测铝基复合材料的TLP扩散焊接头性能是有效的,切实可行的。

Artificial neural network is an effective method to solve the nonlinear relationship between the technics of transient liquid phase （TLP） diffusion bonding non-continuously reinforced aluminium-based composite and its joint performance.Taking relationship between TLP diffusion bonding process of Cu/Al2O3p/6061Al and the anti-shearing strength of joints as the studying example, and considering the data of L9 （33） orthogonal test training and forecasting samples, the BP neural network of having an implicit layer with five nodes was constructed to predict the shearing strength of its joint, the results show that the method is effective, practical and feasible.