Two-Stage Value Iteration for Multi-Leader Tracking under Interactive Nash Equilibrium in Discrete Time
Shangning Liu, Xiaoli Wang, Mu Yan, Yiwen Ma · 2025
For the discrete-time multi-leader system, this paper proposes a two-stage value iteration to fit complex optimal solutions in Bellman equations of multi-leader and realize the tracking target under multi-leader. The two-stage value iteration algorithm adds a sub-iteration that is proposed for obtaining the control strategy. Through this iteration strategy, the optimal control strategy is acquired and the system realizes interactive Nash equilibrium. In this research, we attempt to utilize a critic network for the purpose of fitting the value index as well as an actor network to fit the control strategy. With these two neural networks, we don't need to know the specific system, and the algorithm can be driven by data. At last, simulation results indicate the feasibility of the algorithm.