Artificial intelligence lightning approach early warning method and system based on slicing processing

Abstract

The invention relates to an artificial intelligence lightning proximity early warning method and system based on slicing processing. The method comprises the steps of S1, performing compression processing on an input radar or lightning distribution image; S2, slicing a compressed radar or lightning distribution image according to the forecasting requirement of a certain grid point; S3, establishing a deep learning model, and training an established deep learning model by using the training sets processed in the step S1 and the step S2; and S4, inputting to-be-predicted data into the deep learning model trained in the step S3 for prediction after the to-be-predicted data is processed in the step S1 and the step S2. The service requirements of weather forecast grid point picture input and pixel-level grid point picture output are met, and the calculation amount of an artificial intelligence lightning proximity early warning algorithm is reduced.

Description

technical field [0001] The invention relates to the technical field of lightning impending warning, in particular to an artificial intelligence lightning approaching warning method and system based on slice processing. Background technique [0002] Warning of approaching thunder and lightning. The meteorological forecast generally refers to the forecast of 0-2 hours as impending warning. At present, the mainstream technology of lightning impending warning is to judge and extrapolate the changing trend of radar data and lightning location data to obtain the possibility of radar and lightning data in the future. The position and value, combined with the relationship between the predicted value of radar and lightning and the threshold value of lightning occurrence, or the linear or multivariate multiple equations of radar, lightning and lightning occurrence based on historical process regression, and then the method of obtaining the lightning warning result. [0003] The exist...

AI Technical Summary

Problems solved by technology

[0003] The existing meteorological thunder and lightning approach warning algorithms mainly use linear or simple nonlinear functions in dealing with the relationship equation between radar, lightning and lightning occurrence, and the degree of nonlinearity is low, while the relationship between the occurrence of thunder and lightning and radar and lightning is relatively It is complex. Practice shows that in the field of lightning nowcasting, weak nonlinear functions cannot achieve good forecasting results in some environments (for example, the existing algorithm has good extrapolation forecasting effect in plain areas, and early warning in mountainous areas like Fujian). Ineffective; existing algorithms are biased towards deduction of already generated systems, and weaker than generated forecasts of convective weather systems)

[0004] When the existing mainstream artificial intelligence graphics processing technology trains and predicts radar and lightning distribution pictures, it is difficult to meet the needs of weather forecasting for picture input and picture output (the current mainstream artificial intelligence graphics processing algorithm is suitable for picture input with limited node output)

[0005] When the artificial intelligence algorithm is directly used for the application of lightning impending warning, due to the huge amount of calculation and high requirements for computer resources, the artificial intelligence lightning impending warning algorithm cannot be quickly trained and predicted on general machines, which affects the application technology. Popularization and promotion

Images