航空发动机双冗余架构传感器信息通道故障诊断方法研究

Study of Aircraft Engine Two Redundancy Sensor System Information Channel Structure Fault Diagnosis

传感器数据的高可靠性是航空发动机控制系统可靠工作的基础,故障诊断就十分重要;基于传感器双冗余结构,综合利用神经网络预测和传感器冗余性信息变化判断进行故障诊断是一种新的故障诊断新方法;该方法先用RBF神经网络对传感器输出进行预测,若预测值与输出值发生较大的偏差,进一步考察传感器之间的冗余性信息变化情况来判断传感器是否发生故障,若发生故障,进行故障定位,进而采用对应的诊断策略;仿真实验结果表明该方法能够有效地解决双冗余架构传感器信息通道的故障诊断问题。

The sensor data high reliability is the foundation of the reliable Operation of aircraft engine control system, such as failure diagnosis. This article proposed a new approach of failure diagnosis that is based on sensor two redundancy structure, and composition of neural network prediction and sensor redundant information change to verdict. First of all, this approach utilizes RBF neural network to predict the output from the sensor, if there is a big amount of deviation between predicted value and the actual output, it determines if there is a failure through a further inspection of the variance of redundant information between the sensors, if the failure happens, it determines the possible failure locations, then it carries the corresponding diagnosis strategy.