基于诊断观测器的最优事件触发故障检测

Optimal event-triggered fault detection based on diagnostic observer

提出一种基于诊断观测器的最优事件触发故障检测方法.首先,针对自适应混合事件触发下的远程监控系统,发展一种数值代数结合型的低阶诊断观测器参数矩阵计算方法,以满足残差生成中的Luenberger条件;然后,通过评估事件触发传输误差、干扰和故障对残差信号的影响,构建H∞/H∞型性能指标,并利用互内外分解技术设计最优后置滤波器,以实现对故障灵敏度和对干扰鲁棒性的最优权衡;进一步,为降低传统的预设残差阈值方法的保守性,依据系统初始状态和干扰的有界性,采用基于中心对称多胞体的方法设计残差评价函数及相应的故障决策逻辑来实现故障检测;最后,通过多车跟踪系统的仿真,验证所提方法的有效性.

An optimal event-triggered fault detection method based on diagnostic observer is proposed.Firstly,a low-order diagnostic observer parameter matrix calculation method combining a numerical approach and an algebraic approach is developed to satisfy Luenberger conditions in residual generation for the remote monitoring system which is triggered by adaptive mixed event-triggered mechanism.Then,the H∞/H∞performance index is constructed by evaluating the influence of event-triggered transmission error,disturbance and fault on residual signal,and the optimal post-filter is designed by using the co-inner-outer factorization technology to achieve the optimal trade-off between fault sensitivity and disturbance robustness.Further,in order to reduce the conservatism of the traditional preset residual threshold method,according to the boundedness of disturbance and initial system status,the residual evaluation function and the corresponding fault decision logic are designed based on zonotope to realize fault detection.Finally,the effectiveness of the proposed method is verified by simulation of the multi-vehicle tracking system.