飞行器网络控制系统的多指标鲁棒故障检测

针对存在马尔可夫短时延且状态转移概率部分已知的网络环境,研究了飞行器网络控制系统的多指标鲁棒故障检测问题.传统马尔可夫跳变系统的故障检测方法是将故障、外部输入和未知扰动作为广义输入,采用一个H_∞性能指标来进行故障检测滤波器的设计.由于该指标不具有明确的物理意义,为此需要研究多指标鲁棒故障检测滤波器的设计方法.通过将网络诱导时延描述为一个有限状态马尔可夫链,飞行器网络控制系统被建模为离散有限马尔可夫跳变系统.同时考虑残差对故障的敏感性和对未知扰动及外部输入的鲁棒性,以线性矩阵不等式的形式给出了多指标故障检测滤波器的存在条件和求解方法.最后通过数值算例验证了所提方法的有效性和优越性.

Multi-indexes robust fault detection of aircraft networked control systems

The problem of multi-indexes robust fault detection was investigated for aircraft networked control systems with short Markov delays and partially known transition probabilities. The conventional fault detection method for Markovian jump systems sees fault, input and disturbance as a generalized input, and then a H_∞ index was considered for fault detection filter design. The index has not an explicit physical meaning, so the multi-indexes fault detection filter design method should be studied. The aircraft networked control systems was modeled as discrete, finite-state Markovian jump systems by describing networked induced delays as a finite-state Markov chain. Considering the sensitivity to fault and the robustness against disturbances and inputs on residual, the solvable condition of the multi-indexes fault detection filter and further the solutions were derived by using the linear matrix inequality techniques. The effectiveness and superiority of the proposed method were illustrated by a numerical example.