Linear Optimal Filter for Stochastic Uncertain systems with Multi-loss Packets and Finite-step Delays

This paper studies the linear optimal filter for networked control systems owning non-stationary observations noise. Taking phenomena of multi-loss packets and finite-step delay and uncertain parameters of the networked control systems into consideration, the model of multi-loss packets and finite-step delays are represented by a set of Bernoulli stochastic variables. The parameters uncertainties are depicted by multiplying the white noise sequence. In order to facilitate the derivation of the optimal linear filter, we first perform model transformation. Then the linear optimal filter with linear minimum variance meaning is developed via the projection theory and the innovation analysis technique. A numerical simulation illustrates that the algorithm is effective.