The Application Of Optimization Methods Inthedesign Of Linear Accelerator

Linear accelerator has become a research hotspot in the accelerator field due to its advantage of high flow, high brightness. While for the one-pass acceleration and high RF power dissipation of room temperature structure, the cost of linac become very high when it is used in high energy field. So room temperature linac is usually operated at low energy, low duty factor injector for synchrotron accelerator. Linear accelerator has been required strongly in the fields, such as radioactive ion beam physics, high flux spallation neutron source, Accelerator driven system (ADS), high energy density physics, and fundamental physics. There are many linear acceleration projects all over the world. The running projects include SNS, Linac2, ISAC-I, ISAC-II, CEBAF, the being constructed ones are Spiral2, IFMIF, ESS, FRIB, C-ADS, MYRRHA, Linac4. There are also some planning projects, such as ILC, MEIC, X-ADS. In general, the cost of linac has direct relation with the length of accelerator and the aperture of acceleration structure. The project budget can be reduced by lattice optimization and improvement. For the situation, the optimization design of linear accelerator becomes a very necessary research topic.After the investigation of optimization application in the accelerator field, it is found that for present the optimization problems in the accelerator field are mostly concentrating on lattice of electron synchrotron accelerators. While for the ion linear accelerator optimizations, there are just mating cases in the beam transport line, so the thesis concentrates on the global optimization design of the ion linear accelerator lattice.The main elements of optimization problem are optimization objective function, optimization algorithm and optimization variables. The author applied the method of combination of DAKOTA optimization software and LINREV linac optics design code to do the design of SSC-linac and APF-linac. Through the work of the design optimization o, we studyed the effect on the optimization results by the optimization objective functions, optimization algorithm and optimization variables. The optimization problems in linac are mostly multi-variables, multi-objective problems. In the design of DTL section of SSC-linac, the optimization objective functions are low emittance growth, high transmission. The results shown that the optimization objective functions should be chosen based on physics goal not just mathematic number, and the optimization method should be global intelligent optimization algorithm. In the design of APF-linac, the goals are large acceptance and high acceleration efficiency. From this work, we find that the dimensionality reduction of optimization variables can induce better results. In summary, in the linac optimization problems, simple mathematic optimization will time consuming work and the results are usually not so good. So the linear accelerator optimization should study the three optimization elements and be guided by the physics design principle.Based on the above study, the author proposed his own design optimization method, using smoothness of periodic phase advance as optimization objective, adding other beam parameters to get stable and reliable solution; for the optimization variables, parametric method is used to realize dimensionality reduction of optimization variables. We developed our own design optimization code LinacOpt, which adopts linear matrix mode including space charge effect. LinacOpt can output the optimization goals such as emittance growth, transmission, acceptance, acceleration energy, smoothness of periodic phase advance; the code use particle swarm optimization algorithms to realized the optimization function; for the optimization variables, parametric method is used to realize dimensionality reduction of optimization variables based on the physics rules is adopted to guarantee the automatically lattice design.The code is used to design the APF-linac, compared with the previous method, the results shown that with the same acceleration energy and smaller transverse emittance growth, the longitudinal emittance growth reduce50%, and the iteration number reduce by an order of magnitude. The results prove that the design and optimization function is reliable. The code has the linac lattice optimization design, beam quality optimization, beam parameters matching function.At the end of thesis, the LinacOpt is used to do the lattice optimization design of the Injector II of ADS linac. The results showed that the emittances reduced in both directions. Especially in longitudinal direction, the emittance reduces by60%. The work proves the feasibility and veracity of the global optimization idea and method of dimensionality reduction of optimization variables in further. Major work done during the study period:1. Completion of SSC-linac and APF-linac optimized design DAKOTA with optimization software combined with LINERV software.2. Collaborative-development of our own program LinacOpt, using intelligent optimization algorithms combined to linear accelerator design principles to achieve linear accelerator design optimization.3. Using LinacOpt code to design the ADS Injector II automatically, the results are very good.