超新星遗迹（SNR）是研究“银河系宇宙线起源”的最好侯选体之一，SNR还可以用来研究银河系的演化，具有非常重要的研究意义。SNR作为一个非线性复杂系统与周围的系统甚至整个宇宙系统相互作用。SNR射电辐射的研究可以帮助我们理解该系统磁场、高能电子的行为，进而帮助我们理解整个遗迹系统的演化过程。利用计算机技术，进行合理的物理建模，对超新星遗迹进行三维磁流体数值模拟仿真并将结果可视化，可以帮助天文学家得到一般规律，从而更具体、深刻地理解超新星遗迹的结构、性质以及演化过程。本文提出对超新星遗迹射电形态进行三维磁流体模拟研究，聚焦于模拟对象系统的非线性动力学建模，通过先进的计算机技术实现系统模拟仿真和可视化。对PLUTO三维磁流体模拟算法进行研究；通过改变初始条件进行多次模拟并比较分析；提出新的建模方法与模拟初始化条件，并将模拟结果三维可视化；首次引入强磁场的SNR射电形态演化模拟。实验结果表明，本文的模拟方法得到的射电密度图像与天文实际观测对比分析，可以解释大部分遗迹的射电形态；强磁场中模拟超新星遗迹演化形态帮助理解和解释遗迹射电形态中的多壳层结构。

本文的主要工作和贡献如下：

1.建立超新星遗迹磁流体模拟模型。

通过磁流体力学守恒模型和超新星遗迹的演化模型，结合超新星遗迹的扩散激波加速机制、辐射机制，守恒定律，求解方程，建立本文模拟所使用的遗迹三维磁流体模拟模型。给出了笛卡尔坐标系、极坐标系和球坐标系的磁流体模拟方程，并进行相应的离散化表示。

2.基于PLUTO的磁流体模拟算法研究。

超新星遗迹的模拟遵循时间演化的过程以及相应的守恒定律。PLUTO算法使用Godunov的方法来求解流体力学守恒方程组，并采用高分辨率激波捕捉方案，具有很好的鲁棒性和稳定性。在本文的模拟工作中，PLUTO改进算法具有以下优势：灵活的模块化结构，可以很容易地编码和组合不同的算法来处理不同的物理问题；经过基准测试的数值格式，如黎曼解算器、插值、边界条件，不拘泥于一种方案，可以是多种方案的组合；采用自适应网格技术，提高模拟效率，在同一计算域上解决不同时空尺度的流动特征。    

3.超新星遗迹射电形态的分类及模拟研究。

将超新星遗迹的射电图像进行了统计分类，按照形状特征将其分为七大类。假设遗迹的射电流量密度完全来自同步发射，将模拟结果转换为射电密度图，并根据真实观测的分辨率进行平滑，以便将模拟结果与观测结果进行比较。考虑各种初始条件进行多次模拟，并进行了对比实验，结果显示初始条件的设置对遗迹形态的解释有影响，准确的初始条件可以帮助更好地解释遗迹的形态。                                                                       

4.前身星星风演化结果作为初始条件，进行射电形态的三维磁流体建模、模拟以及可视化。

相关研究先前的模拟所用的初始条件大部分只是采用合理的数值，从而做出定性的研究。本文的工作将星风演化结果作为超新星遗迹模拟的初始条件，模拟了一个以40 km s^-1运动的40M_⊙的前身星爆发后的遗迹演化。结合实际观测，对模拟结果的分析、比较及讨论，证实了大质量前身星的星风对超新星遗迹的射电形态有很大影响，比最初的周围环境影响还大。本文的模拟工作显示，将星风作为初始条件，可以解释大部分遗迹的射电形态，除了多层和不规则遗迹。

5.超新星遗迹在强磁场中演化的三维磁流体模拟。

迄今还没有SNR强磁场相关的模拟工作，本文的工作是开创性的。模拟采用并行计算，在计算机集群提交任务进行模拟工作，并加入了自适应网格技术优化我们的模拟。本文在不同的磁场强度（B）和周边星际介质密度（n）作了三组模拟。第一组，模拟了一个经典SNR在B~9μG，n~0.5cm^-3的普通环境下的演化。第二组，强磁场B~900μG，n~0.5cm^-3，而第三组是致密云的强磁场B~900μG，n~10cm^-3。模拟中，本文采用自适应网格技术，使分辨率达到1024×1024×1024。在并行计算中分块提交任务，缩短模拟时间提高效率。而本文的模拟结果显示，强磁场中演化的超新星遗迹形态比较特殊，能帮助理解和解释遗迹射电形态中的多壳层结构。

Supernova remnant (SNR) is one of the best candidates to study the origin of cosmic rays in the galaxy, which is of great significance. SNR is highly related to massive stars and pulsars, and it can also be used to study the evolution of the galaxy. The study of SNR radio radiation can help us understand the behavior of magnetic field and high-energy electron, and then help us understand the evolution process of the whole relic. As a nonlinear complex system, SNR interacts with the surrounding system and even the whole universe system. Using computer technology and reasonable physical modeling, three-dimensional magnetic fluid numerical simulation of supernova remnants and visualization of the results can help astronomers to obtain general rules, so as to more specifically and deeply understand the structure, properties and evolution process of supernova remnants. In this paper, three-dimensional MHD simulation of radio morphology of supernova remnants is proposed, focusing on the modeling of simulation environment and simulation object; the algorithm of Pluto three-dimensional MHD simulation is studied; new modeling method and simulation initialization conditions are tested, and the simulation process is visualized in 3D; SNR radio Morphology Simulation with strong magnetic field is introduced for the first time. The experimental results show that the radio density images obtained by the simulation method in this paper can explain the radio morphology of most remnants, and the simulation of supernova remnants evolution in strong magnetic field can help to understand and explain the multi shell structure of remnants radio morphology. In this paper, our main work and contributions are listed as follows:

1. Establishment of a supernova remnant MHD simulation model.

Through the conservation model of MHD and the evolution model of supernova remnant, combined with the diffusion shock acceleration mechanism, radiation mechanism and conservation law of supernova remnant, the 3D MHD simulation model used in this paper is established. The MHD simulation equations in Cartesian, polar and spherical coordinates are given and expressed in discrete form.

2. Research on magnetic fluid simulation algorithm based on Pluto.

The simulation of supernova remnants follows the process of time evolution and the corresponding to conservation laws. The Pluto algorithm uses Godunov's method to solve the conservation equations of fluid dynamics, and adopts a high-resolution shock capture scheme. Through the analysis of Pluto code, we can do module research and code test, in our simulation work, Pluto algorithm has the following advantages: flexible module structure, it can easily code and combine different algorithms to deal with different physical problems; after benchmark test, the numerical parameters, such as Riemann solver, interpolation, boundary conditions, does not stick to a scheme. In order to improve the efficiency of simulation, adaptive grid technology is used to solve the flow characteristics of different spatial and temporal scales in the same computing domain.

3.Classification and simulation of radio morphologies of supernova remnants.

The radio images of supernova remnants are statistically classified into seven categories according to their shape characteristics. Assuming that the radio flow density of the relic is entirely from synchronous emission, the simulation results are converted into a radio density map and smoothed according to the resolution of the real observations so that the simulation results can be compared with the observed results. Various initial conditions were considered for simulation and comparative experiments were conducted. The results show that the setting of initial conditions has an influence on the interpretation of the morphology of the relic. Accurate initial conditions can help to better explain the morphology of the relic.

4. The wind evolution results of supernova remnant are used as the initial conditions for the simulation of supernova remnant.

To make qualitative research, we use canonical numerical values in most of the initial conditions. In our work, the results of the stellar wind evolution are used as the initial conditions for the supernova remnant simulation. The evolution of a precursor star with 40 solar mass moving at 40km s^-1 is simulated. The analysis and discussion of the simulation results, combined with the actual observation, confirm our conjecture that the wind of the massive precursor star has a great influence on the radio morphology of the supernova remnant, which may be greater than that from the original surrounding environment. Our simulation work shows that including the stellar wind in the simulation can explain the radio morphology of most of the relics, except for the multi-layer and irregular vestiges.

5. Three dimensional MHD simulation of supernova remnant evolution in strong magnetic field.

Up to now, there is no simulation work related to SNR strong magnetic field. Our work adopts parallel computing, submits tasks in the computer cluster for simulation work, and adds adaptive grid technology to optimize our simulation. We have made three groups of simulation with different magnetic field strength (B) and density (n) of the surrounding interstellar medium. In the first group, we simulate the evolution of a classical SNR in B~9μG, n~0.5cm^-3. The second group uses strong magnetic field B~900μG, n~0.5cm^-3, while the third group uses strong magnetic field B~900μG, n~10cm^-3. In the simulation, we use adaptive mesh technology to achieve a resolution of 1024×1024×1024. In parallel computing, tasks are submitted in blocks to shorten simulation time and improve efficiency. Our simulation results show that the evolution of supernova remnant in strong magnetic field is relatively special, which can help us understand and explain the multi-shell structure in the radio morphology of SNR.