Short-term rainfall prediction method of Beidou navigation system based on convolutional neural network

By combining the convolutional neural network model trained with ERA5 data with Beidou observation data to solve the tropospheric delay and meteorological parameters, the problem of short-term rainfall prediction at sparse or temporary Beidou observation stations is solved, and high-precision short-term rainfall prediction is achieved, which is suitable for scenarios such as outdoor sports and sports events.

CN118780154BActive Publication Date: 2025-09-19HANZHONG YISHENG INTELLIGENT NETWORK TECHNOLOGY CO LTD
2 Cites 0 Cited by

Patent Information

Application Number
CN202410758920.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-13
Publication Date
2025-09-19
Estimated Expiration
2044-06-13

AI Technical Summary

Technical Problem

The existing Beidou-based short-term rainfall prediction method requires a large amount of historical data for modeling and cannot be applied to sparse or temporary Beidou observation stations, resulting in the failure of the prediction algorithm.

Method used

The long-term effective PWV data provided by ERA5 is used to train the convolutional neural network model, and combined with BeiDou observation data to perform short-term rainfall forecasts. The atmospheric precipitation amount is calculated by solving the tropospheric delay and meteorological parameters, and is directly applied to temporary or sparse BeiDou observation stations.

Benefits of technology

It has achieved high-precision short-term rainfall forecasts at temporary or sparse Beidou observation stations, expanded the scope and scenarios of application, and met the needs of outdoor sports and sports events.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention discloses a Beidou navigation system short-term rainfall prediction method based on a convolutional neural network, comprising the following steps: step 1, obtaining historical atmospheric precipitation data and normalizing it; step 2, preliminarily constructing and training a convolutional neural network prediction model; step 3, using the Beidou precise single point positioning method to calculate the total tropospheric delay (ZTD), and using the Sa empirical model to solve the tropospheric static delay (ZHD); step 4, using the relationship between the total tropospheric delay, the tropospheric wet delay, and the tropospheric static delay to solve the tropospheric wet delay; step 5, based on the corresponding relationship between atmospheric precipitation and tropospheric wet delay, calculate the atmospheric precipitation; step 6, using the atmospheric precipitation data as input for prediction. The algorithm can effectively overcome the problem that sparse or temporary Beidou observation stations cannot effectively model short-term rainfall prediction due to the lack of historical data, which leads to the failure of existing prediction algorithms.
Need to check novelty before this filing date? Find Prior Art