航空发动机燃油系统执行机构故障诊断及验证

研究了基于执行机构模型以及发动机逆模型的发动机燃油系统执行机构及其传感器故障诊断方法.基于发动机半物理仿真试验台试验数据建立执行机构小闭环传递函数模型,通过二次多项式拟合将油针位置转换为燃油流量.提出基于自校正在线训练神经网络算法建立发动机逆模型,以离线训练网络参数初始化在线系统,基于阈值更新网络参数,并对学习速率进行自校正,以提高算法的泛化能力及收敛速度.对比执行机构模型输出、发动机逆模型输出与LVDT传感器测量位移换算得到的燃油流量,基于阈值判断故障状态.在T700涡轴发动机半物理仿真试验平台上进行试验,实现了在发动机额定及各种性能退化状态下,执行机构及其传感器漂移和偏置故障的准确诊断及定位,验证了算法的有效性.

Fault Diagnosis and Validation for Aeroengine Fuel System Actuator

A fault diagnosis method for engine fuel system actuator and its sensor based on actuator model and the inverse engine model was studied. The transfer function model of the actuator closed-loop system was built based on the experimental data of the semi physical simulation platform, and the position of metering pin was transformed to fuel flow through quadratic polynomials fitting. An online self-adjustment BP network algorithm was proposed to establish the inverse engine model. In order to enhance the generalization ability and convergence rate of the algorithm, the online training system was initialized by the parameters gotten from the off-line training, the parameter was renewed based on the threshold, and the self-adjustment of the algorithm learning-rate was adopted. Compared the difference from actuator model output, inverse engine model output and the fuel flow by the LVDT sensor transform from the displacement measured, the fault was diagnosed according to the threshold. Semi physical simulation experimented in T700 engine experimental platform. Results show the fault diagnosis and location of the actuator and its sensor in nominal engine and various degradation engine, which shows the validation of the proposed algorithm.