As a key meteorological parameter affecting the characteristics of wind power output

the accuracy of wind speed forecast is directly related to the precision of wind power prediction. With the continuous increase in the proportion of wind power generation in the energy structure

the demand for high-precision wind speed forecasts and high-precision wind power prediction in power grid dispatching and operation is becoming increasingly prominent. In response to the current problem of low wind speed prediction accuracy in numerical weather prediction (NWP) models

this paper proposes a wind speed forecast correction model based on the U-Net deep learning network

using the meteorological variables forecast results output by the NWP model. Furthermore

this paper conducts research on wind power prediction technology based on the corrected results. Taking Northeast China as an example

the results show that the proposed method can reduce the root mean square error of regional wind speed forecasts at different lead times by more than 20%

and increase the correlation coefficient with observed wind speeds by more than 4%

demonstrating a significant advantage compared to other methods. Based on the corrected wind speed forecast results

the accuracy of wind power prediction for wind farms can be increased by up to 2.37 percentage points within a forecast lead time of 1 to 3 days.