A river flow prediction method, device, electronic equipment and storage medium

Abstract

The present application discloses a method and device for predicting river flow, an electronic device and a computer-readable storage medium. The method includes: acquiring satellite remote sensing image data and ground observation station data; using an encoder to process the satellite remote sensing image data Dimensionality reduction processing is used to obtain the flow characteristics of remote sensing rivers; the decoder is used to increase the dimensionality of the ground observation station data to obtain surface observation flow and rainfall characteristics and ground observation weather characteristics; the remote sensing river flow characteristics of the region, the ground observation Flow and rainfall characteristics and surface observation weather characteristics are input into SVM or neural network to obtain river discharge prediction results. It can be seen that the river flow prediction method provided by this application improves the accuracy of river flow prediction by performing dimensionality reduction processing on satellite remote sensing image data and up-dimensional processing on ground observation station data.

Description

technical field [0001] The present application relates to the field of computer technology, and more specifically, to a method and device for predicting river flow, an electronic device, and a computer-readable storage medium. Background technique [0002] At present, the commonly used machine learning methods support vector machine SVM (full name in Chinese: support vector machine model, full name in English: Support Vector Machine) and neural network NN (full name in Chinese: neural network model, full name in English: NeuralNetwork) have been used in the prediction of river flow , where NN includes a variety of neural network classification models. SVM and NN have large prediction errors when the time series of ground observations contains large noise or correlation. When predicting river flow, the measurement error of satellite remote sensing image data observations is large, resulting in observations containing many noise, leading to a decline in the predictive power o...

AI Technical Summary

Problems solved by technology

SVM and NN have large prediction errors when the time series of ground observations contains large noise or correlation. When predicting river flow, the measurement error of satellite remote sensing image data observations is large, resulting in observations containing many Noise, leading to a decline in the predictive power of the model

Images