Grid Wind Speed Prediction Based on DSS-UNet Algorithm
Wind speed forecast has important guiding significance for decision-making and planning in the aviation field.Traditional grid wind speed prediction relies on numerical weather prediction technology.Numerical weather prediction,based on complex mathematical equations,requires high computing power.Given the high hardware requirements of traditional technology and the poor performance of existing numerical weather prediction based on deep learning algorithms,a DSS-UNet algorithm was proposed to predict the next two days of grid wind speed at multiple pressure levels.The MSCSM module of the algorithm fully considers the channel and spatial information of input features.The MSTAM module,improved from the TAM module,was proposed to capture the time information of local and global branches.Experiments were carried out on the open dataset and compared with other algorithms.The experimental results show that the evaluation index of DSS-UNet is better than the compared spatiotemporal prediction algorithms in the experimental area.This algorithm has certain significance for improving the accuracy of weather forecasts and coping with the challenges brought by weather changes.
deep learningattention mechanismwind speed predictionmeteorological grid
万方数据
维普
