The invention discloses a wind farm wind turbine combination process based on wake effect clustering. The method includes acquiring historical winding measuring data, more than one year, of a wind measuring tower corresponding to a wind farm to be measured, and calculating a dominant wind direction of the wind farm to be measured by the wind speed rose diagram analysis method; on the basis of the acquired dominant wind direction of the wind farm to be measured, calculating wake effect on wind speed by using a Larsen model; on the basis of the acquired dominant wind direction of the wind farm to be measured and the acquired wake effect on wind speed, clustering wind farm wind turbines; on the basis of clustering results, calculating wind speed of each wind turbine according to a wake model. The wind farm wind turbine combination process based on wake effect clustering, defects of low stability, poor security and small adaptive range in the prior art can be overcome, and stability is high, security is high and adaptive range is wide.