<p>As wind power continues to be integrated into power systems on a large scale, the effects of active power shortages and reactive power surpluses arising from the bipolar blocking in HVDC systems on the frequency and voltage stability of the receiving-end power grid are becoming increasingly pronounced. Consequently, there is an urgent need for wind power to contribute to frequency and voltage support, thereby strengthening the receiving-end power grid. This paper begins by analyzing the frequency and voltage characteristics of the receiving-end power grid following the blocking incident. It then explores the mechanism by which wind turbines (WTs) can participate in providing frequency and voltage support from the perspective of control methods. Based on frequency security and transient overvoltage limitations, the paper proposes a parameter optimization method for wind power support control. Initially, convolutional neural network (CNN) is employed to extract key features from the sets of frequency and voltage support control parameters. Subsequently, a hybrid model that combines CNN and CatBoost is developed to capture the intrinsic relationships between control parameters and transient indexes. An improved hiking optimization algorithm is then used to optimize the model parameters. By establishing an objective function for transient indexes, the model optimizes frequency and voltage support control parameters that minimize both active and reactive power support. Finally, case studies validate that the parameter optimization method proposed in this paper effectively keeps the frequency and transient overvoltage of wind farms (WFs) within acceptable thresholds.</p>