Flow prediction in the lower Yellow River based on CEEMDAN-BILSTM coupled model

Abstract As one of the important hydrological elements rivers, flow is great significance to development and utilization water resources ecological environment. Based on excellent nonlinear processing capability CEEMDAN advantages BILSTM in time-series data modeling, a coupled CEEMDAN-BILSTM model constructed for prediction, i-month flows from 1951 2016 are used predict 2017 2021. The results show that predicts trend more closely with actual variation, minimum relative error 0.56 maximum 9.48, which maintained within 10%, deterministic coefficients all greater than 0.9, so prediction accuracy high. month i 5 years was picked up by monthly predictions 66 consecutive years, provides new way thinking about river flow.