考虑台风影响的风速多步预测模型

合理认知台风对区域风速预测的影响,对未来风电的最大化利用至关重要.针对台风到来对风速预测影响较大问题,基于多模式集成(ME)的台风数值气象预报信息,提出了一种考虑台风影响的风速多步预测模型.首先,针对台风期间风速数据噪声问题,使用经验小波变换(EWT)对风速历史数据进行解构,基于自适应阈值方法剔除噪声扰动,并重构风速序列信号.然后,采用门控循环单元(GRU)网络对重构后的风速序列进行多步预测,获取不考虑台风影响下的风速预测信息.考虑台风数据的缺乏,在GRU网络基础上引入深度信念网络(DBN)进行台风情况下的修正,以提高考虑台风影响时的风速预测精度.最后,基于中国南部某地实际数据进行算例分析,并与不考虑台风数值预报信息的基础算例进行对比分析.算例分析结果表明,相较不考虑台风影响的基准模型,所提模型可有效降低风速预测误差.

Multi-step Forecasting Model of Wind Speed Considering Influence of Typhoon

The reasonable cognition of the influence of typhoons on regional wind speed forecasting is crucial to the maximum utilization of wind power in the future. Based on the multi-model ensembled numerical weather forecasting information of typhoon, a multi-step wind speed forecasting model considering the influence of typhoons is proposed. In view of the noise of wind speed data during typhoons, empirical wavelet transform (EWT) is used to deconstruct the historical data of wind speed and the noise disturbance is eliminated based on the adaptive threshold method. The sequence signal of wind speed is reconstructed. Then the multi-step prediction of the reconstructed wind speed series is carried out by using gated recurrent unit (GRU) network to obtain the prediction information of wind speed without consideration of the influence of typhoons. For the lack of data during typhoons, deep belief network (DBN) is introduced to realize correction under the condition of typhoon and improve the accuracy of wind speed forecasting considering the influence of typhoons. Finally, a case study is carried out based on the actual data of a weather station in southern China and compared with the fundamental case without consideration of numerical forecasting information of typhoons. The result shows that compared with the baseline model without considering the influence of typhoons, the proposed model can effectively reduce the forecasting errors of wind speed.