Dual control based on rolling optimization

An innovative dual control strategy based on rolling optimization is presented for the parameter uncertainty linear quadratic Gaussian system control problem. By approximately calculating the performance index, the strategy obtains the sub-optimal solution to the original control problem. Adopted Kalman filtering method and by approximating the functional equation of cost-to-go function with respect to the posterior probability, the explicit recursive expression of the original unsolvable functional equation solution is obtained. The coupling between the control and identification is decoupled by the rolling optimization approach.