联邦滤波器容错信息分配方法研究

On Fault-tolerant Information Sharing Scheme for Federated Filters

研究了子系统发生传感器故障时,信息分配系数对故障子系统和无故障子系统性能的影响.在此基础上,提出了根据各个子系统可能发生故障的概率进行信息分配的容错信息分配方法.根据该方法,可降低故障通过反馈重置对其他无故障子系统的污染程度,提高无故障子系统的容错性和备份能力,减少故障信息在重构联邦滤波器中的影响程度和持续时间,从而可使系统快速输出不含故障信息的融合结果.理论分析和实例仿真证明了所提出方法的正确性和有效性.

The effects of the information sharing scheme （ISS） for the federated filter （FF） on the global fusion estimate and the local estimates are analyzed. Then a fault tolerant ISS for the FF running in reset feedback mode is presented. The key to this scheme is that the information factor for each local filter is adjusted according to the probability of arising of the failure in its own corresponding subsystem. This scheme may reduce the contamination from the faulty local filter on healthy local filters via the fusion reset and improve the fault-tolerance and backup capability of the filters. Hence, the reconstructed FF may make a quick recovery to generate global solutions without fault information. Theoretic analysis and simulation show the effectiveness of the method.