Search

This article studies the stabilization control of the Stackelberg game-based system (SGBS) which involves a leader and a follower under continuous time. There are two main contributions: first, the explicit Stackelberg strategy and the optimal performance index under finite horizon are derived based on Riccati equations (REs). Second, the equivalent condition for the SGBS to be stabilizable is obtained, meanwhile, the corresponding Stackelberg strategy and optimal index under this condition are given by REs. Different from the previous works, in the optimization process of the leader, we obtain the linear Stackelberg strategy based on state feedback by applying the matrix maximum principle. Our key technique is solving forward–backward differential equations obtained from the maximum principle.

In this article, we consider the discrete-time rational expectations model with state delays. In particular, the dynamic equation relies both on the history of states and the conditional expectation of the future states. The novelty of this article lies in transforming the solvability of the rational expectations model to that of the forward and backward stochastic difference equations (FBSDEs). The main contribution is to obtain the (unique) solvability and the explicit solutions for the FBSDEs in both finite horizon and infinite horizon. This accordingly gives the explicit solution of the rational expectations model. The key technology is to establish a nonhomogeneous relationship between forward and backward processes of the FBSDEs.

This article aims to solve the multiobjective optimization control problem by investigating the consensus problem of multiagent simultaneous optimization with state and control dependent multiplicative noise. The main contributions of this work are two-fold. First, the optimal consensus control problem is solved by designing the free item of an optimal controller, and the coefficient of the feedback gain is obtained based on analyzing a parameterized generalized algebraic Riccati equation. Second, we derive a new consensus controller that comprises two parts different from the current consensus. One part minimizes a linear quadratic index with a singular control weighting matrix, and the other affords the multiagent systems to achieve mean square consensus.