Optimal Dispatch for Large-Scale Renewable Energy Bases Without Conventional Power Support

In large-scale renewable energy base and HVDC systems, the absence of conventional power support makes commutation failures a significant threat to system security and stability. An optimal dispatch method based on Kolmogorov-Arnold Networks (KAN) is proposed, which considers voltage and frequency dynamic security constraints. First, a time-domain simulation model of the system is developed to generate training data for KAN. The KAN is then designed to estimate the transient overvoltage at the point of common coupling in the renewable energy base after the occurrence of commutation failure. Frequency fluctuation is calculated by frequency nadir prediction model. Thirdly, an optimal dispatch model is formulated to maximize the delivered power of HVDC, which considers both operational and dynamic security constraints. Simulation results demonstrate that the proposed optimal dispatch model enhances system security with an acceptable reduction in delivered power.