损伤检测的经验模态分解法

Damage detection using empirical mode decomposition

用经验模态分解法对几种典型信号的特征进行分析.首先分解出内在模态函数分量,再对模态函数进行希尔伯特变换,得到时频图,由模态分量中突变点的位置来识别损伤发生的时间,而由时频图识别频率的变化.对一单自由度系统在刚度突变和累积疲劳引起的缓慢变化两种情况进行了分析,根据时频图中频率的变化识别出刚度发生突变的时刻及刚度变化的过程和损伤程度.结果表明经验模态分解法是进行损伤检测和时变参数识别比较理想的方法之一.

Several representative signals are analyzed by empirical mode decomposition (HMD) method. The intrinsic mode functions (IMF) are obtained by EMD, and the Hilbert transformation is applied to each intrinsic mode function and time-frequency spectrum is obtained. The damage moment can be detected by the abrupt change in the IMF and the change of frequency can be identified by the time-frequency spectrum. A single degree of freedom suffering from abrupt damage and accumulative damage is studied. The damage moment and change process of stiffness and the damage degree can be estimated by the frequency change in the time-frequency spectrum. The result shows that EMD is one of the effective methods to detect damage and identify time-varying parameters.