Font Size: a A A

The Research And Application Of Short-Time Weather Wind Forecasting Method Based On Deep Learning

Posted on:2021-03-29Degree:MasterType:Thesis
Country:ChinaCandidate:J P LiFull Text:PDF
GTID:2480306470468174Subject:Software engineering
Abstract/Summary:PDF Full Text Request
Weather conditions affect the daily life of people,in the long run the stand or fall of weather conditions are directly or indirectly affect people's production and life,the weather forecast is a vital event,especially in recent years,environmental pollution and global warming problem makes the weather anomaly,natural disasters,such as floods,droughts,landslides,melting glaciers,typhoon,landslide,etc.,weather conditions,such as thunderstorms,hail,dust storms,heavy rain,fog,etc.In the field of meteorology,wind power,as a clean and renewable resource,plays a very important role in the production and life of human beings.In addition,the prediction of meteorological wind power has become an important research field and a hot issue concerned by human beings in the world because of its non-linear and uncontrollable characteristics.Due to the rapid development of data-driven and deep learning,the prediction research of meteorological wind becomes more in-depth.By exploring different new prediction methods,the author hopes to find more meteorological laws and improve the accuracy of wind prediction.This topic to study from the perspective of meteorological data analysis,integrated large-scale time-series data and deep learning technology,in this paper,the main work includes the following:(1)The granger influence test principle and the deep learning Seq2 Seq model were used to mine the causal relationship between meteorological wind characteristics and other meteorological characteristics,and the granger influence score was set to qualitatively and quantitatively describe the granger influence relationship between meteorological wind characteristics and other characteristics.(2)The lstm-based Seq2 Seq wind prediction method studied in this paper was compared with the traditional machine learning prediction model to verify the accuracy and effectiveness of the model by comparing different performance indicators.(3)Based on the short-term meteorological wind prediction model studied in this paper,the short-term meteorological wind prediction system is designed and implemented to make the meteorological wind prediction more convenient and convenient,so as to provide valuable reference for environmental protection,clean energy use and human production and life.
Keywords/Search Tags:Data mining, Wind forecasting, Seq2Seq, The time series
PDF Full Text Request
Related items