工况模态分析的EMD方法

OPERATING MODAL ANALYSIS WITH EMPIRICAL MODE DECOMPOSITION METHOD

EMD(EmpiricalModeDecomposition)算法 1995年由NASA海洋水波实验室提出 ,本质上是一种将时域信号按频率尺度分解的数值算法 ,对于线性时不变系统 ,它可以从时域信号中直接提取具有不同特征时间尺度的内禀模式函数(IMF ,IntrinsicModeFunction) ,分解得到的IMFs之间具有正交性 ,且分解唯一。本文以此为基础 ,将NExT(NaturalExcitationTechnique)方法推广到多点随机激励下的复模态情况 ,对多自由度线性系统实测响应信号的互相关函数进行EMD分解 ,并进而实现模态参数的辨识。

EMD(Empirical Mode Decomposition)method developed by Huang et al working in NASA Laboratory for Hydrospheric Processes/Oceans and Ice Branch is essentially an algorithm that can decompose time signal into IMFs(Intrinsic Mode Function) by different characteristic time scales for linear system.All IMFs are orthogonal and the decomposition is unique.Based on this,NExT(Natural Excitation Technique) is extended to identify the complex structural modes under ambient excitation and to decompose the cross_correlation function of present tested data for multi_degree linear system according to EMD method.