Torus-Event-Based Fault Estimation for Stochastic Nonlinear Systems with Randomly Occurring Saturation and Missing Measurements

This paper is aiming at the fault estimation issue within finite time domain for a type of discrete-time stochastic nonlinear systems. The considered sensor network is affected by randomly occurring sensor saturation and missing measurements. A sensor model is established based on Bernoulli distributions of two known probabilities, which can describe the two types of random incomplete information in a unified framework. Additionally, in an effort to alleviate communication load and maintain data security, the torus-event-based strategy is employed to schedule output data in network control systems. On this basis, a suitable estimator is developed, and sufficient conditions for a fault estimator with reduced conservatism are presented. These conditions guarantee that the dynamic error of the designed estimator meets predetermined performance requirements, and the corresponding fault estimator gains are designed accordingly. Finally, simulation examples are used to testify that the considered estimation algorithm can maintain good estimation performance under torus-event-based mechanisms in complex networks.