2026 Wu Critic Free Policy Iteration Power Frequency

This paper develops a critic-free policy iteration (PI) method for continuous-time linear zero-sum games, aiming to improve the efficiency of policy recovery in these complex systems. The proposed approach directly characterizes the saddle-point policies in the joint policy space, circumventing the conventional reliance on the quadratic value matrix as an iterative variable. A novel policy game Riccati equation (PGRE) is introduced, wherein the unknowns consist solely of policy gains. The solutions of this equation are correlated with the symmetric solutions of the game algebraic Riccati equation, thus ensuring robustness in policy determination. The methodology includes an actor-space approach for executing policy iteration, which guarantees a nonsingular Jacobian at every stabilizing policy. Furthermore, a data-driven algorithm is presented, using a single batch of data to eliminate the need for the value matrix, thereby facilitating an actor-only regression framework. A critical rank condition necessary for unique policy recovery is established, demonstrating the independence of critic identifiability. Experimental validation through a frequency-regulation example in power systems confirms the effectiveness of the approach, while scalability testing highlights significant reductions in both computational and memory requirements.