基于人工神经网络的沿海风速多步预测研究

Research on Application of Artificial Neural Network for Sea Surface Wind Speed Forecasting

基于气象历史观测资料,将长短期记忆网络LSTM方法和Transformer模型结合提出了混合短期风速多步预测模型BLSTM-TRA。以山东半岛南部沿海6个台站为研究区域,通过气象台站观测数据构建数据集。经与2018年ECMWF模式6 h预报结果对比分析,得出如下结论:构建的BLSTM-TRA多步预测模型可大幅度降低风速误差,BLSTM-TRA的1 h单步预测结果和ECMWF预报模式结果对比,其RMSE平均降低了58.9%,MAE平均降低了63.2%;风速误差和大风统计过程分析发现,BLSTM-TRA模型具有一定的抗干扰能力,可以抓住短时大风等敏感信息,对于大风预报结果明显优于ECWMF模式和传统LSTM模型。

Accurate estimation of wind speed is essential for many meteorological applications.A novel short-term wind speed prediction method of the BLSTM-TRA model is proposed by combining the Transformer model and LSTM model.Six stations along the southern coast of the Shandong Peninsula are selected as the research area.After comparing and analyzing the 6 h prediction results of the 2018 ECMWF model,the following conclusions are drawn:The BLSTM-TRA multi-step prediction model can reduce the error of wind speed prediction.Compared with the ECMWF prediction model results,the RMSE and MAE of BLSTM-TRA are decreased by 58.9%and 63.2%on average.The analysis of wind speed error and wind statistical process shows that the BLSTM-TRA model has a certain anti-interference ability and can capture the sensitive information of short-term wind,etc.,which is obviously better than the ECWMF model and traditional LSTM model for wind prediction.