2021 40th Chinese Control Conference (CCC) | 2021
Distributed Filtering for a Class of Saturated Nonlinear Systems with Measurement Outliers
Abstract
This paper addresses a distributed filtering problem for saturated systems over sensor network with measurement outliers, where the multiplicative noises are used to characterize the internal behaviors of system model. A χ2 detection is first constructed in the distributed filter to detect and remove the outliers existing in the measurement outputs caused by some external environment disturbances or malicious attacks. Moreover, the innovation sequence is applied to design filter as a compensation term. The objective of this paper is to find a locally optimal distributed filter such that, for both the measurement outliers and multiplicative noises, an upper bound of filtering error covariance is given and then a proper filter gain is derived to minimize the obtained upper bound matrix with the help of the matrix analysis theory and the probability knowledge. Finally, a numerical experiment is proposed to verify the validity of the designed distributed filtering algorithm.