多传感器采样系统的分布式容错估计

Distributed Fault Tolerant Estimation for Multi-Sensor Sampled Systems

研究了多传感器采样系统在发生一类典型故障情况下的分布式融合估计问题;首先,针对局部传感器,利用Kalman滤波获得的新息进行故障检测;然后在最小方差意义下发展了传感器故障在线递归估计方案;进一步将所获得的估计结果对故障传感器的测量值进行重构,并应用射影定理建立了局部传感器容错更新算法;最后基于线性最小方差融合原则给出了多传感器采样系统的分布式容错估计方案;相比于已有融合估计方法,所提方案不仅能及时检测传感器故障,并且能进一步充分利用故障传感器信息来提高估计精度;数值仿真验证了方法的有效性和优越性。

We deal with the problem of distributed fusion estimation for multi-sensor sampled systems where a kind of typical faults may occur.Firstly,the innovation obtained from Kalman filter is used for fault detection of local sensor.Then an online recursive sensor fault estimation scheme is developed in the sense of minimum variance.Furthermore,using the estimation result to reconstruct the measurement of the faulty sensor,a fault tolerant update algorithm is built with help of projection theory.Based on the linear minimum variance criterion,a distributed fault tolerant estimation scheme is finally developed for multi-sensor sampled systems.Comparing with the existing method,the proposed scheme can not only detect faults in time,but also make full use of information of the faulty sensor to improve the overall estimation accuracy.Simulation results demonstrate the effectiveness and superiority.