Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Radar echo extrapolation model training method based on cyclic dynamic convolutional neural network

A convolutional neural network and radar echo technology, applied in biological neural network models, neural learning methods, neural architecture, etc., can solve the problems of complex echo changes, low utilization of radar data, and strong echoes.

Inactive Publication Date: 2018-11-20
NAT UNIV OF DEFENSE TECH
View PDF3 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] The traditional radar echo extrapolation methods are the centroid tracking method and the tracking radar echoes by correlation (TREC) method based on the maximum correlation coefficient, but the traditional methods have certain deficiencies. Strong and small-scale storm cells are unreliable for forecasting large-scale precipitation; TREC generally regards the echo as linearly changing, but in reality, the echo changes are more complex, and this method is vulnerable to vector field disordered vector interference
In addition, existing methods have a low utilization rate of radar data, while historical radar data contain important features of changes in local weather systems and have high research value

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Radar echo extrapolation model training method based on cyclic dynamic convolutional neural network
  • Radar echo extrapolation model training method based on cyclic dynamic convolutional neural network
  • Radar echo extrapolation model training method based on cyclic dynamic convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0159] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0160] Such as figure 1 As shown, the present invention discloses a radar echo extrapolation model training method based on a circular dynamic convolutional neural network, comprising the following steps:

[0161] Step 1, RDCNN off-line training of cyclic dynamic convolutional neural network: input the training image set, perform data preprocessing on the training image set, obtain the training sample set, design the RDCNN structure, and initialize the network training parameters; use the training sample set to train the RDCNN, input The ordered image sequence of the sequence is forward propagated to obtain a predicted image, the error between the predicted image and the control label is calculated, the weight parameters and bias parameters of the network are updated through backpropagation, and this process is repeated until the predicted result reache...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a radar echo extrapolation model training method based on a cyclic dynamic convolutional neural network. The method comprises a step of RDCNN offline training including the steps of obtaining a training sample set through data preprocessing for a given radar echo image, initializing an RDCNN model, training an RDCNN by using the training sample set, and allowing the RDCNN to converge through a process of calculating an output value through network forward propagation and updating network parameters by backward propagation.

Description

technical field [0001] The invention belongs to the technical field of surface meteorological observation in atmospheric detection, and in particular relates to a radar echo extrapolation model training method based on a circular dynamic convolutional neural network. Background technique [0002] Nowcasting mainly refers to weather forecasting with a high temporal and spatial resolution of 0 to 3 hours, and the main forecasting objects include heavy precipitation, strong wind, hail and other disastrous weather. At present, many forecasting systems use numerical forecasting models, but due to the spin-up delay (spin-up) in numerical forecasting, their short-term nowcasting capabilities are limited. The new generation of Doppler weather radar has high sensitivity and resolution, the spatial resolution of its data can reach 200-1000m, and the time resolution can reach 2-15min. In addition, Doppler weather radar also has a reasonable working mode, comprehensive status monitorin...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/62G06N3/04G06N3/08
CPCG06N3/084G06N3/045G06F18/214
Inventor 李骞施恩马强马烁
Owner NAT UNIV OF DEFENSE TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com