基于CE-EMD分解和BBO-ELM-BP模型的水文站蒸发量预测研究

Prediction of Evaporation at Hydrological Station Based on CE-EMD Decomposition and BBO-ELM-BP Model

为准确掌握大王庙水文站蒸发量数据的变化规律,以该站2009—2021年的蒸发量监测数据为基础,先利用互补式集合经验模态(CE-EMD)开展蒸发量数据的分解处理,将其分解为主趋势分量和周期分量,再通过BBOELM-BP模型实现蒸发量的组合预测,以掌握蒸发量的后期变化规律.分析结果表明:CE-EMD模型能有效实现蒸发量数据的分解处理,并与其他分解模型比较,该模型具有更为有效的分解结果;同时,组合预测结果的相对误差值多在2%~3%,具有较高的预测精度,能有效实现蒸发量的后期预测,并通过不同模型的对比研究,得出该组合预测模型对大样本、长周期的预测效果要优于对小样本、短周期的预测效果,且预测距离对预测结果具有较大影响,主要表现为短距离预测的可信度相对更高.通过本文研究,可有效分析地区的蒸发量变化特征,对掌握其水文规律具有重要意义.

In order to accurately grasp the variation law of the evaporation data of Dawangmiao hydrological station,based on the evaporation monitoring data of the station from 2009 to 2021,the complementary ensemble empirical mode(CE-EMD)is first used to decompose the evaporation data into main trend components and periodic components.Then,the BBO-ELM-BP model is used to realize the combined prediction of evaporation to master the later variation law of evaporation. The analysis results show that the CE-EMD model can effectively decompose the evaporation data,and compared with other decomposition models,this model has more effective decomposition results. At the same time,the relative error values of the combined prediction results are mostly between 2% and 3%,which has high prediction accuracy and can effectively achieve the later prediction of evaporation. Through the comparative study of different models,it is concluded that the prediction effects of the combined prediction model for large samples and long periods are better than that for small samples and short periods,and the prediction distance has a great impact on the prediction results,which is mainly manifested in the relatively higher reliability of the short-distance prediction. Through this study,we can effectively grasp and analyze the evaporation variation characteristics of the region,which is of great significance to grasp its hydrological laws.