Fusion filtering for nonlinear rectangular descriptor systems with Markovian random delays via dynamic event-triggered feedback

This paper focuses on the fusion filtering problem for a class of multi-sensor nonlinear rectangular descriptor systems with random transmission delays by considering the feedback fusion information. The random transmission delays are modeled by homogeneous Markov chains. For the feedback channel, a dynamic event-triggered mechanism is used to determine whether the information from the fusion center can be transmitted to the local estimators side. To facilitate the analysis, the rectangular descriptor systems are transformed into non-descriptor ones through the matrix transformation and full-order transformation methods. Secondly, the received measurements and feedback information are used to design the local filters, where the expression form of filter gains are provided to locally minimize the upper bounds (UBs) of the filtering error covariances (FECs) under given positive parameters. In order to further minimize the UBs of FECs, an optimal selection of the scalar parameters is given to determine the optimal parameters. At the same time, some optimal parameters which are nonlinearly coupled to the optimal gains are solved. Subsequently, a fusion filtering algorithm with dynamic event- triggered feedback is proposed on the basis of the sequential inverse covariance intersection strategy. Finally, a simulation experiment applied to an inverted pendulum controlled by a direct current motor via a gear train illustrates the performance of proposed fusion filtering algorithm.