Robust estimation for discrete time-delay Markov jump systems with sensor non-linearity and missing measurements

This study addresses the ℋ_∞ filtering design issue for a class of time-delay Markov jump system with non-linear characteristics. A stochastic system with sensor saturation and intermittent measurements is considered in the authors study. Random noise depending on state and external-disturbance are also taken into account. A decomposition approach and a bernoulli process are utilised to model the characteristic of sensor saturation and missing measurements, respectively. By transforming the filtering error system into an input-output form, sufficient conditions for the stochastic stability of the system with a prescribed ℋ_∞ level are presented with the help of Scaled Small Gain theorem developed for stochastic systems. Based on the proposed conditions, the rubost filter design approach is proposed. A numerical example is finally provided to demonstrate effectiveness of the proposed approach.