基于小波神经网络的航空发动机传感器故障诊断  

Aeroengine Sensor Fault Diagnosis Based on Wavelet Neural Network

作　　者：白杰[1] 张正 王伟[1] 孙晓楠 BAI Jie;ZHANG Zheng;WANG Wei;SUN Xiaonan(Civil Aviation Civil Aircraft Airworthiness Certification Technology Key Laboratory,Civil Aviation University of China,Tianjin 300300,China)

机构地区：[1]中国民航大学民航民用航空器适航审定技术重点实验室,天津300300

出　　处：《机床与液压》2020年第3期180-186,共7页Machine Tool & Hydraulics

基　　金：大飞机重大专项;天津市自然科学基金资助项目(18JCQNJC05000);中国民航大学研究生科技创新基金资助项目(ZYGH2018018)。

摘　　要：航空发动机传感器故障诊断系统对于保证航空发动机控制系统可靠性和安全性至关重要,针对传统基于发动机模型的传感器故障诊断中存在建模精度不足导致故障诊断存在误诊和漏诊的问题,提出以小波变换和神经网络为基础,使用正常传感器预测故障传感器值。通过对比传感器输出和神经网络预测值的残差来实现传感器的故障诊断,其中神经网络可以在传感器故障后估计出正常的模拟信号代替故障信号供发动机控制系统使用,实现航空发动机控制系统的容错控制;使用改进粒子群优化算法优化BP神经网络的阈值和权值,以提高神经网络诊断和预测信号精度。仿真结果表明:该方法可以有效完成故障诊断,减少漏诊和误诊的发生。Aeroengine sensor fault diagnostic systems are critical to ensuring the reliability and safety of aerospace engine control systems. In view of the traditional engine model-based sensor fault diagnosis,there was a problem of misdiagnosis and missed diagnosis in fault diagnosis due to insufficient modeling accuracy. Based on the wavelet transform and neural network,the normal sensor is used to predict the fault sensor value. The fault diagnosis of the sensor was realized by comparing the residual of the sensor output and the predicted value of the neural network. The neural network could estimate the normal analog signal instead of the fault signal for the engine control system after the sensor failure and realized the fault-tolerant control of the aero-engine control system. The improved particle swarm optimization algorithm( PSOA) was used to optimize the threshold and weight of Back Propagation( BP) neural network to improve the accuracy of neural network diagnosis and prediction. The simulation results show that the method can effectively complete fault diagnosis and reduce the occurrence of missed diagnosis and misdiagnosis.

关 键 词：传感器 小波变换 神经网络 改进粒子群优化算法 容错控制 

分 类 号：V263.6[航空宇航科学与技术—航空宇航制造工程]