Privacy‐Preserving Consensus Control for MASs With Deferred Output Constraint via Reinforcement Learning

ABSTRACT This paper explores a privacy‐preserving adaptive optimal consensus control problem for multi‐agent systems (MASs) with deferred output constraint and external disturbances. A privacy preservation mechanism is developed by introducing a more flexible mask function that converges to the true value within a user‐defined time. The deferred barrier function is formulated and integrated into the optimal backstepping control framework, which ensures the designed optimal controller satisfies the requirements of the deferred output constraint. Meanwhile, the minimization of the cost function and the feasibility of the controller are achieved based on reinforcement learning (RL). Furthermore, the composite updating laws and disturbance observers are designed with prediction errors to provide efficient estimations of the unknown nonlinearities and compound disturbances. Finally, the boundedness of all signals is proved in the closed‐loop system, and two simulation examples confirm the effectiveness of the proposed method.