基于神经网络信息融合的智能故障诊断方法

Fault Diagnosis Technology Based on Neural Network Information Fusion

飞行状态时的飞机舵面故障诊断系统,含有系统和测量噪声及其时变、非线性等特点,采用常规的故障诊断方法很难实现对飞机舵面故障的准确诊断和告警,为了更好的实现对飞机舵面系统的故障诊断,将神经网络信息融合的智能故障诊断方法首次运用到舵面系统故障诊断中。该智能诊断方法应用神经网络的非线性拟合能力扩展舵面相关线位移传感器测量信息,同时采用D-S算法将相关传感器的输出信息进行融合,最后信息融合诊断策略根据这些信息确定出舵面相应的故障类型,从而可以对舵面故障信号进行有效识别和诊断。建立了某机舵面系统故障诊断的数学模型,并利用该模型对提出的智能故障诊断方法进行仿真验证,最后的仿真实验结果表明:该故障诊断结构形式对于舵面常见的故障能够进行识别和告警,诊断效果令人满意。

The plane steering surface fault diagnosis system in flight contains system and measuring noises. Due to their time - changeable and non - linear characteristics, the steering surface fault cannot be precisely diagnosed and alarmed with regular method. In order to improve the fault diagnosis, an intelligent fault diagnosis method based on neural network information fusion is applied in the fault diagnosis of the steering surface system for the first time. The intelligent diagnosis method adopts the nonlinear approach ability of neural network to expand relational linear movement sensor information of the steering surface, synthesizes the output data of relational sensors with the D - S algorithm, and finally the fusion diagnosis strategy ensures the corresponding fault type according to the information, thereby effective identification and diagnosis of the faulty signal are achieved. A mathematical model for the plane steering surface system＇s fault detection and diagnosis is established, with which, the computer simulation of the here mentioned intelligent fault diagnosis method is done. Simulation tests show that the structural form can identify and resume the steering surface＇s normal failure. The effect is satisfactory.