基于SVR的X型发动机传感器故障诊断研究

Research on sensor fault diagnosis for a given turbofan engine based on SVR

利用基于回归型支持向量机(SVR)的诊断方法,设计了某型涡扇发动机传感器常见故障诊断系统,实现了传感器故障隔离与信号重构.通过发动机试车数据对SVR进行训练,以传感器的偏置故障、冲击故障和漂移故障为例,用MATLAB语言进行了计算机仿真验证.结果表明:基于SVR的传感器故障诊断具有精确度高,实时性强的特点,是一种很好的传感器故障诊断方法.

A sensor fault diagnosis system with support vector regression（SVR） for a given turbofan engine was designed.It can detect sensor fault and recover fault-sensor signal.On the basis of training the SVR model with the trial run data of one turbofan engine,the system was simulated in the MATLAB through three different kinds of sensor fault,i.e.offset fault,impact fault and drift fault.The results show that the system has many advantages including high level of precision and good real-time performance.So SVR is an effective method for sensor fault diagnosis.