给出了一种自动识别轴心轨迹形状的新方法,首先将采集到的两相互垂直方向的振动位移信号,经消噪处理拟合为轴心轨迹,然后采用改进后的不变线矩代替传统得HU氏不变面矩,计算其特征向量并作为待识别模式,再计算仿真生成的各种轴心轨迹的特征向量并作为参考模式,通过计算参考模式与待识别模式之间的关联度来对轴心轨迹进行识别。 A new method to automatically identify the shape of the axis trajectory is given. First, the acquired vibration displacement signals in two mutually perpendicular directions are fitted to an axis locus by de-noising, and then the modified invariant moment is used instead Traditional HU constant invariant moment, calculate its eigenvectors and as the pattern to be identified, and then calculate the eigenvectors of the various trajectories generated by the simulation as a reference model, by calculating the correlation between the reference pattern and the pattern to be identified Axis path to identify.