Dynamics analysis and adaptive neural network command filtering excitation control of stochastic power system

New energy sources, such as wind and photovoltaic systems, demonstrate inherent randomness in their power outputs. Additionally, flexible loads, such as electric vehicles at the consumption end further contribute to this variability. These factors result in significant continuous stochastic disturbances on the power system that pose significant threats to the safe and stable operation of the system. Considering continuous stochastic power perturbations, we establish a stochastic differential model of the power system based on Ito stochastic theory. This model analyzes changes in dynamic behavior and oscillation patterns, indicating that stochastic perturbations expand oscillations and reduce the stability boundary. To enhance the system’s security and stability, we propose an adaptive neural network command filter (ANNCF) excitation control method to address stochastic oscillations caused by stochastic power disturbances. Experimental validation using the Real-Time Laboratory (RT-LAB) semi-physical real-time simulation platform shows that the proposed ANNCF excitation method effectively responds to stochastic perturbations, suppresses the stochastic oscillation phenomenon, and significantly improves resistance to stochastic disturbances. Furthermore, this method maintains a superior control effect during sudden power changes and three-phase short circuits, improving the transient stability of the power system.