基于改进联邦滤波的故障隔离与系统重构

为解决故障隔离与系统重构(FIR)问题提出了改进的联邦滤波算法。采用带有备份主系统的联邦滤波结构,避免了由于主系统故障而引发灾难性后果,并针对这一结构提出了分级联邦滤波FIR策略,保证在任何一个子系统故障的情况下的滤波连续性和滤波结构完整,并提高了后续故障的检测能力。在融合算法上提出了分时融合反馈的联邦滤波信息分配方式,大大降低全局被污染的可能性,并提高了系统对渐变故障的检测能力。将该算法应用于多传感器组合导航系统中,仿真结果表明在任何一个子系统故障情况下,滤波器都能提供连续、准确的导航信息。

Fault isolation and recovery based on improved federated filter

An improved federated filtering method is presented for fault isolation and recovery(FIR).A double master system federated filtering structure is adopted to avoid the fatal effect caused from a master system failure.According to this structures a FIR strategy based on hiberarchy federated filter is proposed.If one sensor of the filter is failed to work,it can also produce a continuous estimate.The subsequent fault detection can be performed as well.A time-sharing feedback federated filter is presented based on information sharing principle to reduce the possibility of global contamination and improve the capability of detecting the gradually changed fault.The application of the algorithm on the multisensor integrated navigation system proves that the filter could produce accurate and continuous navigation information,whichever subsystem goes wrong.