• • Distributed consensus ACE discovery enables each regulation unit to communicate only with adjacent units, eliminating the single-point failure sensitivity of centralized AGC and reducing computational complexity from O(n²) to O(n) per iteration, which is critical for maintaining frequency stability when communication infrastructure is partially damaged during disasters.
• • Heterogeneous unit coordination is achieved through independent PI controllers tuned to each unit's dynamic response: gas turbines and diesel generators with slow ramp rates (governed by time constants T_GT1–T_GT4 and T_DG1–T_DG2) are coordinated with fast-response wind (T_WT), photovoltaic (T_PV), and fuel cell (T_FC) units, preventing the fast units from being curtailed by centralized allocation.
• • The latter-half regulation strategy adjusts slow-response unit output to release fast-response unit capacity, reserving frequency regulation margin for subsequent cycles; this is analogous to state-of-charge management in storage systems and directly addresses the capacity depletion problem that occurs when fast units are continuously dispatched under sustained disturbances.
• • The proposed method avoids the retraining requirement of multi-agent reinforcement learning approaches (e.g., DDPG, Q-learning) when extreme disasters cause generation unit failure or topology changes, providing a decisive operational advantage where retraining latency of minutes to hours would be unacceptable for real-time frequency control.