Abstract: Aiming at the significant characteristics of randomness and volatility of a wind farm, there exist problems of steady-state error and slow response speed under conventional PI control based on MMC flexible HVDC technology. WNN (wavelet neural network)is applied to the outer ring of voltage control due to its characteristic of approximating any nonlinear function. PSCAD/EMTDC and Matlab/Simulink are used to build system of electrical part and WNN control part respectively. Simulation results show that WNN control strategy has good dynamic characteristics and better guiding effect for large-scale renewable energy grid.

Key words: mountain wind farm, flexible HVDC transmission system, PI control, WNN