To address the issue of power fluctuations caused by disturbances and parameter variations in the pitch control of wind turbines，a pitch angle control strategy based on the fusion of an improved genetic algorithm and predictive control （PID+ImGA+MPC） is proposed.This method introduces a Cauchy-uniform distribution hybrid crossover and adaptive mutation mechanism to enhance global search capability，and combines ARIMA wind speed prediction with model predictive control to achieve feedforward adjustment of the pitch angle. By deeply integrating the improved genetic algorithm，ARIMA wind speed prediction，and MPC，an integrated "prediction-optimization-feedforward" pitch control framework is formed，which compensates for the shortcomings of existing strategies，such as insufficient utilization of predictive information and reliance on offline optimization for parameter tuning.Hardware‑in‑the‑loop verification is conducted using a pitch control system intelligent loading test bench based on Bladed.The results show that the error between simulation and measured results is controlled within ±3%，verifying the engineering feasibility and real‑time performance of the strategy.In simulations on a 1.5 mW doubly‑fed wind turbine model，under step wind conditions，the proposed strategy reduces the power overshoot from 4.025% to 0.69% and shortens the settling time from 3.66 s to 2.2 s；under turbulent wind conditions，the standard deviation of output power is reduced by 12.40% compared with conventional PID.This study can provide a theoretical basis and engineering reference for the stable control of wind turbines under high wind power penetration conditions.