Fusion Estimation Of Multi-rate Uncertain Systems With Unknown Sensor Interference

The fusion estimation problems for uncertain systems with unknown sensor disturbance are widely involved in application fields, such as information processing, engineering control and fault diagnosis. The immeasurable interference in the sensor is referred to as unknown sensor disturbance(i.e., unknown inputs). With the advent of ‘Internet+' era, the development of networked control systems(NCSs) will be more deeply. However,as the scale of NCSs increase ceaselessly, the complexity of the system is increasing gradually, some problems have cropped up. The phenomenon of missing measurements is inevitable in data transmission due to the poor network capacity and the limited bandwidth. In addition, the uncertainties issue arising from multiplicative noises cannot be neglected. They will lead to production loss if we wave those above disturbances aside or those disturbances are failed to be detected and isolated. Nowadays, for the sake of getting more comprehensive information to improve system performance, multi-rate multi-sensor systems have become a hotspot research in industries, because of single sensor system and single-rate system can no longer meet the requirements. Taking into account the issues above, it's necessary to study fusion estimation problems for multirate uncertain systems with unknown sensor disturbance. The main contents are as follows:The research of this dissertation is based on the linear unbiased minimum variance criterion and the optimal fusion algorithm(in the linear minimum variance sense).For multi-sensor asynchronous uniform sampling systems with unknown sensor disturbance and missing measurements simultaneously, the Kalman-like local filters independent of unknown sensor disturbance at the observation sampling points and the state update points are respectively presented. Furthermore, two kinds of suboptimal fusion filters that avoid calculation of local estimation error cross-covariance matrices are designed: SCI fusion filter & the suboptimal fusion filter weighted by matrices.For multi-sensor asynchronous uniform sampling systems with unknown sensor disturbance and missing measurements simultaneously, the Kalman-like local filter without relying on unknown disturbance at the state update points is directly proposed. Then, the estimation error cross-covariance matrices between any two sensor subsystems are derived in three different situations. At last, the distributed fusion filter with higher estimation accuracy is given.For multi-sensor asynchronous uniform sampling systems with unknown sensor disturbance, missing measurements and multiplicative noises simultaneously, the Kalman-like local filter without relying on unknown sensor disturbance at the state update points is directly proposed. Multiplicative noises are described by Gaussian white noise with zero mean. Then, the local estimation error cross-covariance matrices are derived in all sampling cases. Finally, the distributed optimal and suboptimal fusion filters weighted by matrices are given.