基于EMD方法与RBF神经网络的结构损伤检测

Structural damage detection based on EMD method and RBF neural network

针对结构损伤检测中损伤的识别、定位以及程度的标定这三个独立并按一定先后顺序进行的检测过程,提出了一种能将以上三者同时进行的联合检测方法。该方法首先利用经验模态分解(EMD)方法将三层钢筋混凝土剪切型结构在各种损伤工况下的顶层地震作用加速度响应分解为若干固有模态函数(IMF)分量,然后以此IMF分量和未经EMD分解的原始加速度响应数据来构造损伤标识量,作为特征参数依次输入到径向基函数神经网络(RBFNN)中进行损伤检测。给出了应用此方法的具体步骤,通过仿真实验证明了利用该方法进行结构损伤一次检测的可行性和有效性,结果表明,由加速度响应经EMD分解而得到的IMF分量输入到RBFNN中能够更为精确地一次检测出结构所有损伤信息,并且RBFNN在结构损伤损度大时具有更好的检测效果。

Aiming at three independent detection processes carried out according to certain order of sequence, identification, orientation and severity estimation of structural damage, a combination detection method that can carry out the above three processes simultaneously was put forward. Firstly, this method adopted the empirical mode decomposition（EMD） method to decompose the top-story seismic acceleration response of a threestory shear reinforced concrete building under several damage conditions into some intrinsic mode function （IMF） components. Then the damage identification vectors were constituted by using IMF components and original acceleration responses not decomposed by EMD method. Those damage identification vectors were used as input characteristic parameters for radial basic function neural network （RBFNN） to detect structural damage. The concrete step of this combination detection method was given, and the numerical simulation experiment demonstrated its feasibility and validity.