BP Neural Network-Enhanced Active Disturbance Rejection Control for Inertial Synchronous Control of Permanent Magnet Synchronous Wind Turbines

Inertial synchronous control (ISC) enables gridforming operation while maintaining the control logic of active power regulation on the generator side and DC-link voltage control on the grid side for permanent magnet wind turbines (PMWTs). However, existing ISC strategies suffer from DC-link voltage fluctuations induced by power oscillations, leading to voltage sensing deviations and reduced stability margins. To address these issues, a BP neural network-enhanced active disturbance rejection control (BP-ADRC) strategy for DC-link voltage stabilization is proposed. First, an ISC model is established, and its small-signal stability is analyzed to clarify the mechanism of DC-link voltage fluctuations caused by power oscillations. Second, BP-ADRC is integrated into the generator-side speed loop control. A cost function is constructed with the objectives of mitigating wind speed and actual speed oscillations, thereby achieving robust DC-link voltage disturbance rejection. Finally, simulation validation is conducted on the Simulink platform. Results demonstrate that the proposed strategy effectively suppresses DC-link voltage fluctuations while accurately tracking grid frequency variations, significantly improving grid frequency sensing accuracy and grid-connected stability.