Research On Severe Convection Weather Recognition Technology Based On Intelligent Computing

Strong convective weather has strong destructive power and obvious suddenness.The occurrence of severe convective weather often poses a huge threat to the safety of life and property of the country and people and the guarantee of military operations.Therefore,this article has carried out corresponding research on the production of more sophisticated strong convective weather warning products and the improvement of the accuracy of strong convective weather identification and prediction.First,this paper builds a severe convective weather knowledge map based on text data.It is mainly divided into three parts: severe convective weather corpus construction,knowledge extraction and knowledge map storage.For the problem of severe convective weather knowledge map,some documents with the theme of “severe convection” are firstly labeled.Combining word embedding technology with neural network Bi-RNN and CRF model,a severe convective weather knowledge extraction model is constructed.Finally,the extracted knowledge is converted into a triple form and imported into the graph database Neo4 j,which completes the storage and visualization of knowledge,and provides data and information support for the study of severe convective weather.Secondly,a middle-level radial convergence(MARC)recognition algorithm is proposed.The radar base data interpolation in polar coordinates is transformed into profile grid data.At the same time,template matching is used to realize the identification of severe convergence points and obtain the parameters of convergence strength and position.The centering line of the convergence area is used to correct the convergence strength.Effectively overcome the underestimation of MARC due to radial velocity.Finally,through the expression of thunderstorm on radar data,the classification algorithm of thunderstorm cloud cluster identification is designed.Firstly,the screening of thunderstorm process was completed by combining lightning location data.Secondly,the parameters of the dual-polarization radar are analyzed.The temperature,reflectivity factor,differential reflectivity factor and differential propagation phase constant are selected as the identification factors of thunderstorm.Finally,the two-step discrimination method is used to complete the classification and classification of thunderstorms.