Real-time monthly runoff forecasting method based on deep learning model

Abstract

The invention provides a real-time monthly runoff forecasting method based on a deep learning model, and the method comprises the steps: 1, collecting forecasting factors based on historical information and future meteorological information, and determining the longest delay of the influence of the early monthly runoff on the forecast monthly according to the autocorrelation analysis of the monthly runoff in the historical period of a drainage basin; 2, performing normalization processing on forecast factors and monthly runoff data in a training period, and automatically screening the forecast factors by adopting an LASSO regression method based on an embedded thought; 3, clustering the training period sample set by adopting a K-means clustering method based on a division thought, and dividing samples into K classes which do not coincide with each other; 4, calculating the distance between the forecasting factor vector of the verification set and the clustering center of the K training sets, finding the nearest training set, and then training a combined deep learning forecasting model combining the convolutional neural network and the gating circulation unit network by using the data set; and 5, carrying out real-time correction on the forecast residual error by adopting an autoregressive moving average model.

Description

technical field [0001] The invention belongs to the technical field of hydrological forecasting, and in particular relates to a real-time monthly runoff forecasting method based on a deep learning model. [0002] technical background [0003] Monthly runoff forecasting is one of the important engineering and technical problems in the field of hydrology. It can not only provide information support for solving the uncoordinated natural water and man-made water use, but also guide the development and management of water resources in the basin. One of the premise and basis of natural disasters. As a weakly correlated and highly complex nonlinear dynamical system, the monthly runoff process has high requirements for the construction of forecast models. [0004] Generally speaking, monthly runoff forecasting models can be divided into two types: process-driven and data-driven. The process-driven model, also known as the physical cause analysis method, needs to rely on a hydrologi...

AI Technical Summary

Problems solved by technology

[0005] However, there are three major problems in the existing monthly runoff forecasting methods based on deep learning: (1) The source of the predictor is single, only the observation data in the historical period are considered, and the future meteorological information predicted by the numerical forecast product is not considered as the predictor; (2) The screening method of predictors is single, and most of the current studies use filtering methods (such as Pearson correlation coefficient, mutual information coefficient, etc.) to screen predictors. (3) The forecast model is single and fails to consider the temporal heterogeneity of monthly runoff, especially for runoff in flood season and non-flood season, the use of a unified model lacks consideration

In addition, there is still a certain gap between the accuracy of the monthly runoff forecast output by the existing deep learning model and the actual demand

Images