基于故障因子函数的鲁棒联邦滤波算法研究

Study of a Robust Federated Filtering Algorithm Based on Fault Factor Function

针对联邦滤波器子系统传感器故障,提出了一种基于状态递推器的滑动残差检验法,并在故障检测函数的基础上,定义了故障因子函数,用于联邦滤波器的信息分配。根据新的算法,联邦滤波器的信息分配系数可根据故障因子函数进行自动调整,以减少故障通过全局重置对其他无故障子系统的污染程度,从而减少了故障信息的影响程度和持续时间,提高了故障隔离后联邦滤波器的快速恢复能力。仿真结果表明该算法是有效的。

This paper firstly proposes an improved moving residual test with high fault detection sensitivity based on the state propagator for the federated filter. Then the fault factor function is defined to share the system information of the federated filter. With an analysis of the sensor failure＇s influence on the good subsystems with different information coefficients, an adaptive information sharing algorithm is presented to improve the recovery capability of the federated filter by heightening the robustness of the good subsystems. By this algorithm, the information sharing coefficients are adaptively adjusted according to the fault factors, and the faulty sensor＇s contamination on the good sensors via the global fusion reset is reduced. So after fault isolation, the fault information still in good subsystems will only last a very short time in the regenerated federated filter. The simulation shows this method is effective.