关于航空发动机传感器故障信号优化诊断研究

Fault Signal Optimizing Diagnosis of Sensors in Aeroengine

研究航空发动机传感器故障诊断问题。由于传感器输出样本有限并且存在噪声,传统的诊断方法无法消除噪声的影响。为了消除噪声的影响、提高传感器故障诊断的准确性,提出了一种采用模糊支持向量机(fuzzy support vector machine,FSVM)的航空发动机传感器故障信号优化诊断方法。首先该方法运用FSVM和传感器时间、空间上的冗余信息建立传感器预测模型,然后利用预测值与实测值两两之间的残差进行故障诊断,最后进行仿真验证。仿真结果表明,改进方法可以在样本存在噪声的情况下对传感器进行准确有效的故障诊断,提高航空发动机的可靠性。

Fault diagnosis of sensors in aeroengine was researched in the paper. Because the fault samples of engine sensor are limited and have noises,traditional fault diagnosis methods cannot eliminate the effect of noises. For reliably finding sensor faults,a method of fault signal optimizing diagnosis for sensors in aeroengine control system was presented based on FSVM. First,the fault prediction model of sensors was built using FSVM and redundancy information of the sensors. Then,the residual sequences were generated from predicted value and measured value. Finally,the method was verified by simulation experiment. Experiment results show that the FSVM algorithm can effectively detect the fault of the sensor system in case of sample having noise,and improve the reliability of aeroengine.