With the integration of large-scale wind power clusters into the power system, wind farms are increasingly involved in reactive power regulation of the power grid. Based on this, an upper level optimization model is first established to minimize the active power loss and voltage deviation in the power system. Subsequently, the model is solved to reduce system network loss and achieve safe and energy-saving operation of the power grid. Additionally, a detailed analysis of the wind power cluster is conducted at a lower level, estimating its reactive power potential. Taking the reactive power potential of the wind farm as the objective function and the reactive power output of each wind turbine unit as the optimization variable, this study coordinates the reactive power output within the wind farm cluster and employs an improved whale optimization algorithm to solve this bi-level optimization model. Finally, the proposed optimization strategy and algorithm are verified through case studies to effectively reduce network losses and voltage deviation in the power system while maximizing utilization of wind power′s reactive power potential.