具有参数摄动的时变系统的间歇故障检测

Detection of Intermittent Faults for Time-varying Systems Subject to Parameter Perturbations

针对一类具有参数摄动的随机时变系统,研究了其鲁棒间歇传感器故障检测问题。存在模型不确定性情况下,根据最小二乘准则设计了无偏最小方差状态估计器和相应的残差生成器。随后设计了基于最大似然比的残差评价函数,并利用其统计特性设定了故障检测阈值。该故障检测方法为递推方法且不依赖模型不确定性结构,因此,适用于实时在线应用。仿真实验结果表明,所提方法的有效性。

The robust detection problem of intermittent sensor faults for a class of stochastic uncertain systems subject to parameter perturbations is investigated. In the presence of model uncertainties, the unbiased least mean-square state estimator and corresponding residual generator are designed. Subsequently, the residual evaluation function is presented based on maximum likelihood ratio. According to its statistical properties, the fault detection threshold setting method is also given. The proposed fault detection method is recursive and independent of uncertainty structure, thus suitable for real-time online applications. Finally, a simulation example is provided to illustrate the effectiveness and applicability of our proposed method.