Collaborative Simulation And Optimization Based On Python

Due to the complex and highly nonlinear interaction processes inside the electrovacuum devices,theoretical analysis is usually difficult,while experimental studies are used with high processing costs,long iteration cycles,and internal physical quantities that are not easily observed.Therefore,in order to better carry out the research of electric vacuum devices,PIC simulation using computer is a common research tool,and with the help of PIC simulation,the changes of physical quantities inside the device that are not easy to observe can be observed very clearly,providing reference for theoretical analysis and final design,saving a lot of human and material resources and speeding up the development cycle.In addition,the PIC simulation process inside the electric vacuum device is very complex and requires consideration of many parameters,such as voltage,magnetic field,etc.,and also deals with microscopic collisions of particles in the process of motion,which leads to a very large amount of calculation of the corresponding PIC simulation.Therefore,the optimization of parameters in PIC simulation was often achieved by manual trial-and-error or parameter scanning in the past,which cost a lot of time and human resources.Optimization theory is a good solution to this problem.When the number of parameters is large,the optimization algorithm can be used for fast optimization in the PIC simulation process and improve the optimization efficiency in the design process.Therefore,this thesis takes relativistic magnetron as a reference,and develops and designs a Python-based PIC co-simulation optimization system by studying the modeling ideas and modeling methods of CST and CHIPIC,which are commonly used simulation software,and implementing optimization algorithms in Python.For CST,based on the external implementation of the optimization framework,we add the working state determination in the optimization process to ensure that CST in PIC simulation and optimization will not have self-excited oscillations due to grid dissection or energy nonconservation due to non-convergence of calculation results;for CHIPIC,we introduce thread pool technology and network communication to improve the computational efficiency of CHIPIC on the basis of the optimization function.By introducing thread pooling technology and network communication for CHIPIC,the computational efficiency of CHIPIC can be improved based on the optimization function.The system can realize or improve the functions of CST and CHIPIC in PIC simulation.By customizing the algorithm parameters,the researcher no longer needs to change the parameters repeatedly by hand for trial and error,and can observe the optimization process in real time.Based on the designed co-simulation optimization system,the six-cavity and eightcavity relativistic magnetrons are simulated and optimized using this optimization system,and the feasibility of the co-simulation system is verified by comparing the results before and after the optimization.In addition,the co-simulation optimization platform in this thesis is not only applicable to the design and optimization of relativistic magnetrons,but also to the design and optimization of general electric vacuum devices.