基于动态模态分解的输电杆塔松动检测

Transmission tower looseness detection based on dynamic mode decomposition

为了识别输电杆塔的松动状态,提出分段高阶动态模态分解(segment high-order dynamic mode decomposition,SHDMD)的检测方法。为减小振动耦合对检测的影响,利用三轴加速度和三轴角速度构造时间-空间矩阵,从中提取准确的方向振型作为松动特征,将时间-空间矩阵在时间维度上划分为若干子矩阵,对每个子矩阵进行空间维度扩展,避免DMD分解时得到错误结果,对扩展后的子矩阵进行DMD分解,得到不同时段的振动模态,利用稳定图筛选出不同时段共有的模态作为真实模态,提取与模态一一对应的方向振型。建立灰色关联检测模型,通过计算方向振型的几何特征关联度,识别当前松动状态。模拟杆塔试验与真实杆塔试验结果证明,所提方法能够很好地实现输电杆塔松动位置与松动程度的识别。

Here,to identify loose state of transmission towers,a segmented high-order dynamic mode decomposition(SHDMD)detection method was proposed.To reduce effects of vibration coupling on detection,a time-space matrix was constructed using tri-axial acceleration and tri-axial angular velocity to extract correct directional vibration mode shapes as loosening features.Firstly,the time-space matrix was divided into several sub-matrices in time dimension,and each sub-matrix was expanded in spatial dimension to avoid incorrect results during performing DMD.Then,DMD was performed for expanded sub-matrices to obtain vibration modes in different time intervals,Stability maps were used to screen out common modes in different time intervals as actual modes,and extract directional vibration modes corresponding to each actual mode one by one.Finally,a grey correlation detection model was established to identify the current loose state by calculating geometric feature correlation degrees of directional vibration modes.The results of simulated tower tests and actual tower tests showed that the proposed method can effectively identify looseness positions and levels for transmission towers.