中文摘要：

针对传统载荷大小识别在应用上的局限性,利用分形理论中的关联维数,对单跨转子系统的响应振动信号进行定量特征提取,实现转子系统的载荷类型识别。在正弦、稳态、线性、冲击4种类型载荷作用下,对不同测点的不同方向的实测振动信号,利用改进阈值处理函数进行降噪,通过作图法选定几个重要参数,计算降噪后信号的关联维数,对载荷的类型进行判断识别。结果表明：在4种类型载荷下,对振动信号进行关联维数计算,可以实现定量的特征提取,关联维数的结果具有明显的分段性,能够实现4种不同载荷类型的有效识别。

英文摘要：

In view of the limitation of traditional load identification more in size recagnition,correlation dimension of fractal theory was used to extract quantitative characteristics of vibration signals from single-span rotor system to identify the load type of system. Under the four types of sinusoidal,steady,linear and impulse loads ,the vibration signals of different directions acquired in different positions were denoised by an improved threshold processing functions,and were selected by images to ascertain several important parameters and calculated correlation dimensions to identify the load types. The results show that correlation dimensions of vibration signals under the four types of load can realize the quantitative feature extraction and identify four types of load because of segmented features.