基于局部方法的余度惯性组件故障诊断

Fault detection for inertial measurement unit based on local approach

介绍了局部方法(local approach)原理及其在随机系统故障诊断中的应用问题,讨论了用于故障诊断的样本数目对检测性能的影响,指出通过增加样本数目,可以大大增加局部方法检测小故障的能力,并说明样本数为1时,广义似然比(GLR,Generalized LikelihoodRatio)方法和局部方法的判决函数相同.为了验证所提出方法的有效性,将局部方法用于沿正十二面体对称侧面配置的冗余微机械惯性组件(IMU,InertialMeasurementUnit)故障检测,实验结果表明文中的理论分析是符合实际情况的.

The application of local approach for fault detection in stochastic dynamic systems was presented. It was demonstrated that the selection of samples numbers plays an important role in improving the performance of the local approach. And it was pointed out that by enlarging the number of samples, the ability of the technique to detect small fault will be enhanced significantly. It was shown that when the sample size n = 1, decision function of the generalized likelihood ratio （GLR） method is the same as that of the local approach. In order to illustrate the efficiency of the proposed method, the local approach was applied to detect fault of the redundancy micromechanical inertial measurement unit （IMU） with a dodecahedron configuration, and the experiment results confirm findings of the theory analysis.