从2017年开始，我们开发了基于隐空间条件生成模型的三维像素卷积循环网

Since 2017, we have developed deep learning technologies such as three-dimensional pixel convolutional recurrent network models and adversarial generative network models based on latent space condition generation models to conduct short-term nowcasting of severe convective storms and classified severe convective weather, forming a system A complete set of advanced classified severe convective weather monitoring, short-term forecasting and short-term nowcasting technology systems that combine the application of physical laws and mechanisms with artificial intelligence.

Since 2017, we have developed deep learning technologies such as 3D pixel convolutional recurrent network models based on hidden space conditional generation models and adversarial generation network models to perform short-term proximity forecasting of severe convective storms and classified severe convective weather. We have formed a complete set of advanced monitoring, short-term forecasting, and short-term proximity forecasting technology systems that combine physical laws and artificial intelligence to classify severe convective weather.

Since 2017, we have developed deep learning technologies such as three-dimensional pixel convolution cyclic network model and confrontation generation network model based on hidden space condition generation model to carry out short-term approaching forecast of severe convective storms and classified severe convective weather, and formed a set of advanced monitoring, short-term forecasting and short-term approaching forecasting technology systems for classified severe convective weather by combining physical laws and artificial intelligence.