基于局部投影算法的齿轮故障声信号特征提取

局部投影算法采用延时坐标将时间序列进行相重构，在高维的相空间上采用局部投影的方法将相空间分解成正交的子空间，通过子空间中吸引子特性的不同来分离时序中的背景信号和弱特征信号分量。提出将局部投影算法用于设备故障声信号的降噪，通过齿轮故障信号的特征提取实验证实该方法用于识别设备故障的有效性。

Gear fault diagnosis acoustic signal feature abstract based on local projective noise reduction

Time series is reconstructed in phase spaces using delay coordinates.In the high dimensional phase space,the local projective algorithm is used to separate phase spaces into orthogonal sub-spaces,ground signal and weak feature signal in the time series is separated by the attractors which indicate different characters in the sub-spaces.The noise reduction of gear fault acoustic signal based on local projective method is proposed,through the experiment for the abstracting of feature signal of gear with fault,it is proved that the algorithm has an excellent validation on identifying machinery fault signals.