Elman网络在某型自动驾驶仪飞控盒多故障诊断中的仿真研究

目前采用传统故障诊断方法对复杂系统进行故障诊断时，还存在许多缺点，尤其在多故障同时发生的情况下，传统故障诊断方法难以准确定位故障点。为此，基于Elman神经网络，结合故障字典法，设计了一种基于Elman网络的综合故障诊断方法。为验证该方法的准确性，以某型飞机自动驾驶仪飞控盒的几种主要故障为诊断源，仿真研究了所设计的综合故障诊断方法。仿真结果表明，与传统故障诊断方法相比，所设计的基于Elman网络的综合故障诊断方法在进行多故障诊断中对故障的准确定位是有效的和快速的。

Simulation Study of Multi-Fault Diagnosis Method of Flying-controlled Box in the Autopilot Based on Elman Neural Network

There are many disadvantages of dealing with faults by means of the traditional fault diagnosis methods in the complex systems at present,especially many faults occurring at the same time.A kind of integration fault diagnosis method is designed,combining with trouble dictionary method,based on Elman neural network.In order to validate validity of the designed fault diagnosis method,simulation researches are conducted with the major failure of certain kind aircraft autopilot's flying-controlled box as the example.The simulation results show that this designed diagnosis method,compared with the traditional fault diagnosis methods,is effective and available to solve many faults taking place at one time.