一种基于支持向量回归的发动机振动监控方法

A Monitoring Method of Aero-engine Vibration Based on the SVR

基于某型飞机FDR记录的飞行数据,采用支持向量回归(SVR)方法,建立了航空发动机振动的初始状态辨识模型。通过监控模型输出值与实际值之间偏差的大小,确定发动机是否发生异常振动。根据实际的振动故障样本给出了故障监控的偏差参考门限。现有的振动故障样本验证了本方法的有效性。

Based on the flight data recorded by FDR equipped on a certain aircraft, the model of inertial phase of aero- engine vibration is established by Support Vector Regression （SVR）. The abnormal vibration is detected by the difference between the output of model and the real value recorded. According to real malfunction samples, a referenced criterion of malfunction monitoring is given. The existing malfunction samples proved this method to be effective.