Fault Detection for Linear Discrete Time‐varying Systems with Intermittent Observations and Quantization Errors
Fault Detection for Linear Discrete Time‐varying Systems with Intermittent Observations and Quantization Errors
复制标题
DOI：
10.1002/asjc.1041
复制
发表时间：
2016-01
影响因子：
2.4
通讯作者：
Yueyang Li;Shuai Liu;Zhonghua Wang
中科院分区：
文献类型：
--
作者：
Yueyang Li;Shuai Liu;Zhonghua Wang
研究了具有多个间歇观测值和量化误差的线性离散时变系统的故障检测问题。引入一组独立同分布的随机变量作为观测序列的指标来描述多个间歇观测。所测量的观测值由对数型量化器量化。我们的重点是构建一个基于观测器的故障检测滤波器（FDF），以识别故障，尽管多个测量数据包丢失和量化不准确。通过定义广义输入输出算子，将FD问题转化为一个双目标优化框架，使得随机H∞/H∞或H−/H∞性能指标最大化.基于伴随算子的优化方法分别推导了两种情况下的概率/指标依赖和概率依赖的解析解。一个是指标是在线已知的，而另一个是指标在每个时刻都不可用。最后通过一个算例验证了该方法的有效性和适用性。
This paper deals with the fault detection (FD) problem for linear discrete time‐varying (LDTV) systems subject to multiple intermittent observations and quantization errors. A set of independent identical distributed random variables are introduced as the indicators of the observation sequences to describe the multiple intermittent measurements. The measured observation is quantized by a logarithmic type quantizer. Our focus is to construct an observer‐based fault detection filter (FDF) to recognize the fault in spite of multiple measurement packet dropouts and quantization inaccuracy. By defining generalized input‐to‐output operators, the FD problem is formulated into a two‐objective optimization framework such that stochastic H∞/H∞ or H−/H∞ performance index is maximized. Probability/indicators‐dependent and probability‐dependent analytical solutions are respectively derived by virtue of an adjoint operator based optimization approach for two cases. One is that the indicators are on‐line known while the other one is that the indicators are not available at each time instant. An illustrative example is employed to demonstrate the effectiveness and applicability of the proposed approach.