A second-order neurodynamic optimization based automatic generation control method for power systems

The increasing integration of renewable energy sources (RES) into power systems poses significant challenges such as frequency instability and unpredictable power fluctuations. To address these challenges and improve the tractability of automatic generation control (AGC) problems with nonlinear dynamics while maintaining reliable solution performance, this paper develops a Lagrangian-dual second-order neurodynamic optimizer (LDSONO). The proposed scheme embeds second-order unit dynamics within the optimization mechanism, allowing the algorithm to explicitly incorporate the physical behavior of AGC while achieving stable convergence. The AGC framework is first reformulated with regulation objectives and dynamic constraints, after which the neurodynamic evolution equations of LDSONO are constructed. Theoretical analysis establishes the existence of equilibrium solutions and verifies the stability of the resulting control dynamics. Case studies on a 15-unit system demonstrate fast frequency recovery and coordinated regulation under consecutive disturbances, with improved mileage performance relative to benchmark AGC strategies. Additional tests under complex communication conditions and system scales validate robustness and computational scalability, indicating that LDSONO offers an efficient and practical solution for AGC applications.