River water level prediction method considering time lag effect

Abstract

The invention discloses a river water level prediction method considering a time lag effect, and the method comprises the following steps: 1, determining lag values of upstream and downstream water levels based on a distribution lag model; 2, decomposing water level data based on a VMD model; 3, reconstructing a VMD decomposition component to obtain a component combination; and 4, training and predicting the reconstruction component combination based on the BP neural network. The distribution lag model is a regression model based on a time sequence, can reflect the lag effect between variables, and provides an effective method for determining lag values of upstream and downstream water levels. The variational mode decomposition method is an analysis method for processing nonlinear and non-stationary signals, and can perform linear and stationary processing according to the characteristics of the signals. The BP neural network is a multi-layer mapping network for carrying out weight training on a nonlinear function and has good nonlinear fitting. The method combines the three models, gives full play to respective characteristics and advantages, achieves the prediction and forecasting of the downstream water level based on the upstream water level, and obtains good prediction precision.

Description

technical field [0001] The invention relates to the technical field of water resources management, in particular to a river water level prediction method integrating a distributed lagging model, a variational mode decomposition and a BP neural network. Background technique [0002] Water level is the most intuitive factor reflecting the water regime in a basin, and it is also an important indicator for hydrological forecasting and water resource evaluation. Water level information is helpful to understand the dynamic changes of river water level, and has important reference value for flood control and drought relief and water resource scheduling. Prediction of river water level is conducive to mastering water level information and flood conditions, and early warning work can be done in advance to minimize losses caused by flood disasters. At the same time, water level monitoring equipment is expensive, and the construction and management of hydrological monitoring stations ...

AI Technical Summary

Problems solved by technology

But one of the disadvantages of wavelet decomposition is that different mother wavelets have different decomposition properties

Images