Multi-meteorological-factor mode forecast temperature correction method and system based on deep learning

Abstract

The invention discloses a multi-meteorological-factor mode forecast temperature correction method and system based on deep learning. According to the method, the memory and analysis capability of datain a period of time and the key point concentration capability of an attention mechanism can be trained by using LSTM; a basic universal model is obtained, the process that a large amount of statistical analysis and processing need to be carried out on data in the past is avoided, and therefore the method has high speed and an excellent correction effect. Moreover, due to the fact that the data are inseparable in low dimension, the method conducts high-dimension mapping on the data, and items of higher dimension and mutual relation of the features are obtained. By utilizing different characteristics of the items and clustering, modeling can be carried out in a more targeted manner, and temperatures with different characteristics are corrected by utilizing different models. And finally, the designed comprehensive indexes are used for evaluating the correction effect and selecting the model, so that the singleness and incomparability of the indexes are avoided, and the applicability ofthe model is improved.

Description

technical field [0001] The present invention relates to the field of meteorology and the field of artificial intelligence technology, and more specifically relates to a method and system for correcting temperature in multi-meteorological factor model forecast based on deep learning. Background technique [0002] At present, meteorology is closely related to human activities. The accuracy of weather forecasts greatly affects the military, people's livelihood, economy and other fields. Extreme weather changes will even destroy the human living environment. The grid meteorological element forecast is a method of dividing the area into grid points according to the predetermined range, and the meteorological element forecast is made with the grid point as the unit. At present, there is a 5km grid point meteorological element forecasting business. The grid point forecast of refined meteorological elements is a kind of high-resolution forecast in the grid point forecast. The foreca...

AI Technical Summary

Problems solved by technology

It solves the problem of poor processing ability of a large amount of data in machine learning, large noise and easy over-fitting in the correction of model forecast temperature based on deep learning method, and the problem of insufficient feature mining in deep learning, and does not separate data with different characteristics modeling problem

Images