Accurate wind speed prediction is the basis of wind power forecasting

and the evaluation of wind prediction is one of the effective ways to improve wind power forecasting. Based on the mesoscale numerical weather research and forecast model（WRF）

and combined with the numerical forecast results of the Northwest Power Grid

the applicability of six kinds of planetary boundary layer parameterization schemes was studied by using quantitative analysis indicators such as threat score

correlation coefficient and root mean square error. The results show that the YSU（Yonsei University）scheme is the optimal boundary layer scheme for improving the wind field of the Northwest Power Grid

the highest average correlation coefficient of the surface zonal wind is 0.87

the lowest root mean square error is only 1.0 m/s

the forecast effect of the surface meridional wind

and 850 hPa and 500 hPa height fields

and 500 hPa wind speed is the optimal. This study points out the direction for the improvement of the forecast effect of wind field（and other meteorological elements） for Northwest Power Grid.