数据驱动的陀螺随机游走故障诊断研究

Research on Random Walk Fault Diagnosis of Gyro Based on Data-driven Method

针对工程中陀螺工作期间出现随机游走超差而导致的随机游走故障,提出了一种基于数据驱动的故障诊断方法。该方法基于陀螺实测无故障脉冲数据,利用Allan方差对无故障数据进行噪声特性分析与建模,并在此基础上进行故障模拟,然后利用小波包理论对故障样本进行时频域分析,结合信息熵理论提取样本的故障特征,再利用神经网络构造分类器,将故障特征与故障模式形成映射,进行故障模式的分类。本文以某型号激光陀螺为例展开研究,仿真结果表明,该方法简单、不依赖于模型、诊断快且诊断效果好。

For the random walk fault caused by random walk being out of tolerance during working hours of gyro,a data-driven fault diagnosis method is proposed. The noise characteristics based on Allan variance are firstly analyzed and modeled,and the failure samples on the model are simulated. Then,wavelet packet and information entropy theory are used to extract the fault feature vectors of data,and neural network is used to establish classifier to realize classification of fault modes. A laser gyro is taken as an example in this paper,and the simulation results show that the proposed method is simple,independent of model,fast and effective.