中文摘要：

针对旋转机械中出现的大量局部故障现象，采用基于模型的故障诊断方法提取剩余振动量，利用模态扩展推测转子系统各节点瞬态振动，针对模型诊断中等效力出现的发散情况，提出通过最小二乘法利用挠曲线确定单一周期和准周期故障发生位置。提出利用瞬时能量判别瞬时故障力大小的方法，从而诊断出系统中故障严重程度。实现了仅利用少量传感器即可对转子系统进行精确故障位置和故障力大小的诊断。对转子系统的碰摩故障进行数值仿真，并在转子试验台进行试验验证本方法的正确性。

英文摘要：

Model based fault identification techniques can be used to diagnose local faults in rotating machinery; The equivalent loads that represent fault forces can be used to identify the fault location. But in some cases the equivalent loads smeare over many nodes in using this method, and make it hard to identify accurate fault location and transient fault force. The fault location is identified using least squares fitting approach by the system＇s vibration shape when the fault signal is periodic or quasi-periodic. And after the fault location is ascertained, the transient fault forces can be identified by transient residual vibration using simple matrix multiplications and additions. The fault force location and magnitude of the rotor system can be detected precisely by using only a few sensors. Numerical simulations on rotor to stator rub and experiment in testrig are used to test the method, which can prove the efficiency of the method.