基于多源数据模态分解算法的气象短时预测算法OA

Meteorological phenomena are affected by a variety of physical processes,and the interaction between these processes is complex,which makes the meteorological data show nonlinear and non-stationary characteristics.It is difficult to capture the high-frequency instantaneous changes and low-frequency periodic changes,resulting in the low accuracy of meteorological short-term prediction.Therefore,a meteorological short-term prediction algorithm based on multi-source data mode decomposition algorithm is proposed.The complete ensemble empirical mode decomposition with adaptive noise is used to decompose the multi-source meteorological time series data once to determine the high-frequency signal and low-frequency signal.The most complex high-frequency signal is decomposed twice by using variational mode decomposition to generate several eigenmode func-tion components,so as to obtain the high-frequency instantaneous changes and low-frequency periodic changes.The deep deter-ministic cyclic jump state network is constructed based on the obtained eigenmode function components,and the prediction re-sults of each network are superimposed to achieve short-term accurate prediction of meteorological.The experimental results show that the proposed algorithm has low Theil inequality coefficient,high protocol index,good performance of Taylor chart,and accurate meteorological short-term prediction results.