Multi-Classification of Rainfall Weather Based on Deep Learning-Mod

Rain is one of the common weather conditions in the meteorological field, and at the same time it will have an important impact on people's daily life, industry and agricultural development. Rainfall is related to physical parameters of weather conditions, such as temperature, saturated water pressure, water vapor density, temperature and humidity. This paper constructs a rainfall multi-classification model based on physical quantities and deep learning model DBNs. Firstly, to resolve the problem of uneven distribution of the samples of the original data, some rainfall samples were synthesized using the SMOTE algorithm. Secondly, the observation data encrypted hourly by the automatic ground monitoring station and physical quantities commonly used in weather forecast analysis are used to construct a deep learning model containing a Gauss Boltzmann machine to extract the intrinsic characteristics of the original data. Finally, a multi-class identification model of DBNs rainfall weather was realized. Experiments show that compared with other methods, the model can more accurately identify rainfall weather, and the POD, FAR and CSI indicators all perform well.