The working conditions determine that the turbo-shaft engine is easy to instantaneously imbalance in the transition state,which is easy to cause the rotor to to encounter rubbing faults.Rotor rubbing faults can cause sudden changes in some statistical characteristic parameters.Based on ternary search tree algorithm,the rotor local rubbing fault monitoring method was proposed.The vibration signals of rotor rubbing state and normal state of the turbine casing were simulated based on frequency-doubled envelope curve and rotor speed curve of the turbo-shaft engine rotor vibration signal.The vibration signal frequency spectrum analysis was carried out,and the kurtosis index,margin index and total amount were extracted to identify the rotor rubbing faults.The research results show that the fundamental frequency of the rotor is easy to highlight the fault characteristics;the rotor rubbing fault can be well identified by using the rubbing monitoring method.