Short-term Wind Power Forecasting Based on Meteorological Forecast Data of Multiple Isobaric Surfaces

Considering that more than 90% of water vapor is concentrated in the troposphere, and many weather phenomena such as cloud, fog and rain occur in the middle and lower troposphere, a short-term wind power prediction method based on meteorological data of multiple isobaric surfaces in the middle and lower troposphere is proposed. In view of the fact that the power accuracy depends greatly on the accuracy of wind speed prediction, and the current wind speed prediction in wind farms has some problems such as phase lag and system deviation, the decoupling method of wind power conversion model is considered, which firstly corrects the wind speed prediction error and then establishes the wind power conversion model. In detail, based on the multi-layer meteorological forecast data, the optimal feature combination optimization model is established with the maximum average accuracy as the objective function. On this basis, a wind speed correction model based on Conv1D was constructed. Finally, the wind speed - power conversion model is constructed by five kinds of fluctuation process classification, and the corrected wind speed is converted into power. Experimental results show that the proposed method can obtain better accuracy than using correlation coefficients to select feature combinations or only using surface data for modeling.