结构损伤识别的混合神经网络方法

On hybrid neural network method for structural damage identification

在损伤位置和程度的诊断中分别构造各自相适当的网络和识别样本形成混合神经网络系统,并选择不同的输入参数对结构损伤的位置和程度进行了识别,针对一个损伤钢梁进行了数值仿真,从而证明了混合神经网络两步检测法的可行性。

The paper establishes the respectively suitable network and the identification samples to the hybrid neural network system from the diagnosis for damage position and the levels,selects the various input parameter to identify the structural damage position and the levels,and undertakes the numerical simulation for the damaged steel beam,so as to prove the feasibility of the hybrid neural network two-step checking method.