Hydrological forecasting method based on deep learning and transfer learning

Abstract

The invention relates to the technical field of hydrological forecasting, in particular to a hydrological forecasting method based on deep learning and transfer learning, and the method comprises the steps: screening main contribution features; training a long and short-term memory model by using historical data of the main contribution features; constructing a weight distribution model based on the difference of the contribution rates of different main contribution characteristics and the contribution rate attenuation of the corresponding main contribution characteristics in a preset time interval; optimizing a loss function of the long-short-term memory model according to a weight difference between a calculation weight obtained by the weight distribution model and a training weight obtained by the trained long-short-term memory model to obtain a hydrological forecasting model; and correcting the hydrological forecasting model according to the error between the forecasting result and the actual result under different orders of magnitude to obtain an optimized hydrological forecasting model corresponding to each order of magnitude. According to the method, the loss function of the model can be optimized by using the influence between different hydrological characteristics, and the training effect of the network model and the learning ability of the newest data are improved.

Description

technical field [0001] The invention relates to the technical field of hydrological forecasting, in particular to a hydrological forecasting method based on deep learning and migration learning. Background technique [0002] Hydrological forecast is a qualitative or quantitative forecast of the hydrological situation in a certain period of time in the future based on previous and current hydrological and meteorological information. Hydrological forecasting is an important basic water conservancy work and non-engineering measures for flood control, which directly serve the rational utilization and protection of water resources, the construction and management of water conservancy projects, and industrial and agricultural production. Hydrological data is continuous, periodic and seasonal. Because hydrological data changes with time and is affected by factors such as periodic and seasonal changes, changes in hydrological data generally follow certain laws. [0003] At present,...

AI Technical Summary

Problems solved by technology

[0004] The traditional physical environment analysis method is easily affected by noise data or partial mutation data, the prediction accuracy of the model is low, and it is easy to fall into the local optimal problem; the possible problem of the deep learning prediction method is that the hydrological time series data involved in the training are required. , the model tends to fall into the problem of overfitting

Images