All-weather solar radiation remote sensing forecasting method under deep learning support

Abstract

The invention relates to an all-weather solar radiation remote sensing forecasting method under deep learning support. The method comprises the following steps: S1, acquiring time sequence multispectral remote sensing data; s2, performing data preprocessing on the obtained time sequence multispectral remote sensing data to obtain a preprocessed time sequence multispectral remote sensing data set; s3, inputting the obtained preprocessed time sequence multispectral remote sensing data set into a ConvLSTM network model to obtain instantaneous solar short-wave radiation prediction data or multi-time scale solar short-wave radiation prediction; the ConvLSTM network model is a deep learning network model which faces remote sensing images and is combined with a space attention mechanism. According to the forecasting method provided by the invention, the spatial characteristics and the time sequence characteristics of the solar radiation are completely considered, a spatial attention mechanism is added in the deep neural network, and mutation of the solar radiation in the nature is more effectively captured.

Description

technical field [0001] The invention belongs to the technical field of remote sensing forecasting, in particular to an all-weather solar radiation remote sensing forecasting method supported by deep learning. Background technique [0002] In the field of solar photovoltaic forecasting (solar radiation forecasting), accurately predicting the trend of solar radiation changes and predicting the value of solar radiation in advance is of great significance for new energy industries such as photovoltaic power generation peak forecasting and photovoltaic power station site selection. With the continuous development of deep neural network methods, some neural network models have been applied to the prediction of solar radiation and have shown good performance. However, the training and testing data that traditional deep neural network models rely on when predicting solar radiation are mostly site data, that is, point data at a single location, and the true value referenced by tradit...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the above-mentioned problems in the prior art, the present invention provides an all-weather solar radiation remote sensing forecasting method supported by deep learning, which breaks through the inability of the traditional neural network to take into account both the accuracy of the absolute value of solar radiation and the distribution of spatial and temporal characteristics in the process of solar shortwave radiation prediction. The problem

Images