Solar Wind Research And MHD Simulation Based On Big Data Technology

The sun is experiencing the nuclear fusion at any moment,the temperature of solar corona is more than 1,000,000?,in the case of which helium and hydrogen in the sun are completely ionized into protons and electrons,the energy released from the corona expansion can cause the accelerating motion of these particles,charged particles constantly break away from the sun gravitation and radiate around,and these high-speed particle flows released from the corona hole are just named as solar wind.Based on near earth observation,the mean velocity of solar wind is up to 200~800km/s.It is divided into two categories,one is turbulent solar wind,which is a high-energy and high-speed flow of particles released due to corona expansion during the solar activity,with shorter explosion time,strong injection of particles and occurrence of many earth hazard events;the other is a quiet solar wind,which is a flow of particles thrown out of the calm corona area of the sun,with lower flow rate,longer lasting time and steadier state,10 or less protons /m3,long time of disturbance on the geomagnetism,more obvious regularity and remarkable influences on the weather situations in the solar-terrestrial space.In terms of researches on the solar-terrestrial space,spatial physics and spatial meteorology,etc,the research on the simulation of solar wind presents important reference values and significance.Big data of the solar wind studied in this paper come form monitoring data of CE-1(Chang'e-1 satellite),Fengyun meteorological satellites,Helios(running orbit 0.3AU-1AU),Ulysses(running orbit 1AU-5AU),ACE(Advanced Composition Explorer),WIND and SOHO,etc.,as well as research results of others.Firstly,the solar activities are researched,the observation properties of solar wind at the place of 1AU are analyzed and summarized in this paper;secondly,a neural network technology is applied for classification and prediction researches on the big data of solar wind;thirdly,MHD simulation model of solar wind is analyzed and established from solar wind particle motion model,magnetic fluid property and simulation algorithm;finally,a MHD simulation cloud platform of solar wind is constructed,Ansys FLUNT software is applied to realize three scenarios of magnetic fluid simulation of solar wind on the simulation cloud platform and complete the analysis of such simulation result.Research results and innovation points in this paper are as follows:(1)A clustering analysis for big data of solar wind based on SOM neutral network is proposed.In order to quickly acquire key data from mass big data of solar wind,this paper hereby brings forward a clustering analysis for big data of solar wind based on SOM neutral network.First of all,all data are directly read in the data pool,then boundary functions of data are found according to data characteristics and SOM neutral network method is applied to realize the data classification.As shown in the experiment,this method can accurately identify the big data of solar wind,such as density of protons,velocity of solar wind,Dst indicator and F10.7 indicator,etc.Furthermore,this method simplifies the acquisition mode of key data of solar wind and improves the analysis and research efficiency of solar wind data.(2)A method for solar wind correlation factor analysis and velocity prediction is proposed.According to the law of energy conservation,the energy carried by the solar wind is proportional to its speed.During the realization of the solar wind simulation,the velocity is an important parameter.The solar wind is interfered by many factors during transmission,some of these influences are known but some are unknown.Based on research on a large number of observation data,solar wind velocity correlation factor analysis based on grey theory and solar wind velocity prediction method based on RBF neutral network are proposed in this paper.The correlation analysis result shows that the relations of Kp indicator,AE indicator and sunspot number are closely related to the solar wind velocity change;during the predictive research,RBF neutral network is applied to predict the solar wind velocity,which is compared with blurry Hidden Markov Model prediction method,this method in this paper optimizes the threshold value and weight value between two layers of neutral networks,a decision tree algorithm is used for data after training and optimization so as to realize further attribute reduction and rule selection and guarantee the comprehensive efficiency and detection precision of rapid detection on the solar wind velocity data.(3)A prediction method for solar wind correlation factors based on BP neutral network is proposed.The core of research on solar wind simulation aims to comprehensively understand the solar wind phenomenon and realize the prediction on the phenomenon,Kp indicator,AE indicator and sunspot number are closely related to the solar wind phenomenon.On the basis of analysis of solar wind correlation data,a solar wind correlation factor prediction method based on BP neutral network is proposed in this paper,BP neutral network technology is used to construct a solar wind correlation factor prediction model and realize the predictive research on Kp indicator,AE indicator and sunspot number.The experiment shows that the prediction result of this method is more accurate,with a smaller error.(4)An improved PAC non-local means denoising method based on solar wind simulation image is proposed.During the simulation,solar wind may suffer from different types of interferences,which results in the decline in quality of the simulation image and affects the use and analysis of this image.Aiming at PCA-NLM denoising method which easily loses image textures,an improved PCA non-local means denoising method based on description of texture characteristics is put forward in this paper to restore the true expression of this image.From PSNR and SSIM indicators and denoising vision,this method better retains details of the solar wind image,with better acceleration and overall denoising effect.(5)A MHD numerical model of solar wind is proposed and its simulation is realized.The simulation of solar wind by magnetic-fluid method is a complex process,and the realization of this simulation is dependent on the design of simulation platform,the construction of mathematical model and the selection of simulation algorithm.In order to solve the problem of bottlenecks during simulation,a model of solar wind simulation platform based on cloud computing technology is hereby presented in this paper,which solves the bottlenecks during the magnetic-fluid simulation of solar wind.A MHD numerical model of solar wind is brought forward during mathematical research of the simulation model to describe the motion state of solar wind by differential system of equations,research the motion equations of solar wind particles in the electric and magnetic fields,construct the solar wind magnetic-fluid control equation and realize the simulation of three scenarios of magnetic fluids of solar wind.In order to achieve a better simulation effect,a contrastive simulation research is carried out and then the simulation result conforms to some known satellite monitoring data and empirical models.