The problem of robust predictive control is investigated for stochastic singular systems with polytopic uncertainty,
and a design method of robust predictive controller based on off-line state observer is proposed. By constructing the Lyapunov
function with the error terms, using multidimensional Itˆo formula and LMI methods, the “min-max” stochastic programming
problem is converted into the solvability problem of a group of LMIs, and then the expression and sufficient conditions for the
existence of the control law are given. It is proved that the stochastic admissible property of the closed-loop singular systems
is guaranteed by the initial feasible solutions of the optimization problem. A simulation example illustrates the effectiveness
of the proposed method.