基于集成学习的黄河未控区径流预测研究

Research on Runoff Prediction of Uncontrolled Areas of the Yellow River Based on Ensemble Learning

以黄河未控区径流预测为研究目标,利用伊洛河、沁河流域降雨、植被覆盖、土地利用、社会经济数据,结合数据特征提取和关联分析方法,实现对结构化和非结构化要素数据的筛选、清洗、插补、格式转换等数据预处理及初步分析.基于集成学习中的极端梯度提升树(XGBoost)算法设计并构建了大数据驱动的黄河未控区径流智能预测模型.并以2003年洪水过程作为验证数据,与传统水文模型HBV模型进行效果比较.为黄河流域干支流未控区径流预报提供借鉴和参考.

Taking the runoff prediction of the uncontrolled area of the Yellow River as the research goal.The use of rainfall,vegetation coverage,land use,and socio-economic data in the Yiluo and Qinhe river basins,combined with data feature extraction and correlation analysis methods,enables the screening,cleaning,interpolation,and format conversion of structured and unstructured element data pre-processing and preliminary analysis.Based on the XGBoost algorithm in ensemble learning,a big data-driven runoff intelligent prediction model for the Yellow River uncontrolled area was designed and constructed.The 2003 flood process was used as verification data to compare the effect with the traditional hydrological model HBV model.It provides a reference for the runoff forecast of the uncontrolled areas of the main and tributaries of the Yellow River Basin.