基于数据驱动的主燃油计量装置故障诊断

Data-driven fault diagnosis of main fuel metering device

关键词:  航空发动机  燃油计量活门  传感器  故障诊断  模型降阶  QAR数据

Key words:  Aero-engine  Fuel metering valve  Sensor  Fault diagnosis  Model order reduction  QAR data

摘要:

为建立主燃油计量装置的故障诊断方法，提出一种基于快速存取记录器（QAR）数据的燃油计量参数估计模型建模方法，并结合液压机械控制装置工作原理，给出了基于燃油计量参数残差设计的主燃油计量装置故障诊断方案。设计基于系统辨识的燃油计量活门位置（FMVP）估计模型，采取降阶处理策略以提高模型的估计精度和动态响应速度，在此基础上给出基于多项式拟合的燃油流量（FF）估计模型。结果表明：所建立的FMVP模型估计残差不超过±2%，FF估计模型残差不超过±5%；该方案能够有效地诊断出燃油计量活门故障、线性可变差动位移传感器故障和热式流量计故障。该方案结构简单，不需要大量的调试技术，提高了故障检测精度，具有工程应用前景。

Abstract:

To establish a fault diagnosis method for the main fuel metering device， a Quick Access Recorder （QAR） data-based modeling approach was proposed for estimating fuel metering parameters. A fault diagnosis scheme for the main fuel metering device was given by combining fuel metering parameter residuals with the working principle of hydraulic mechanical control device. A fuel metering valve position （FMVP） estimation model was designed using system identification， whose estimation accuracy and dynamic response speed were improved using a model order reduction strategy. Furthermore， a fuel flow （FF） estimation model was presented by polynomial fitting. The results showed that the estimated residuals of the established FMVP model do not exceed ±2%， and the residuals of the FF model do not exceed ±5%. The proposed method can effectively diagnose faults in the fuel metering valve， the linear variable differential transformer displacement sensor， and the thermal flowmeter. The proposed scheme has a simple structure and does not require extensive debugging techniques， and offers higher detection accuracy as well as promising engineering application prospects.