Application Of EPICS And Research Of Beam-tuning Methods In The 320kv Heavy Ion Multidisciplinary Research Facility

This thesis summarizes the design concept of accelerator control system,takes the 320 kV Heavy Ion Multidisciplinary Research Facility(hereinafter referred to as HIMRF)as the application object,analyzes control requirements of each controlled equipment in detail,and a brand new control system was designed and implemented according to the three-layer standard architecture utilizing the design concept proposed including: control network,high voltage platform and beam line control.Specifically,the front-end control layer adopts high performance programmable logic controller(PLC)for local control of all analog and digital I/O interfaced devices,and used the bulk data transmission protocol and Modbus TCP protocol through Ethernet for data exchange with the integration layer,and serial port servers were used to convert the serial interface into Ethernet to communicate with the integration layer.The integration layer is developed based on the EPICS software framework and finite state machine design model,and all the front-end controlled devices are uniformly published as channel access process variables.Based on the channel access,the application layer used Control System Studio to develop the graphical operating software.At present,the new control system is stable and reliable,greatly reducing the failure rate of HIMRF.Based on the new implemented control system,this thesis introduces the current situation of manual beam tuning at HIMRF,analyzes the reasons for the difficulties in manual beam tuning,introduces the Bayesian Optimization(BO)method and Differential Evolution(DE)method,and puts forward the considerations of automatic beam tuning according to the existing conditions of HIMRF.In order to test the practical effects of these two optimization methods,online automatic beam tuning optimization experiments were carried out on the beam transport line of NO.3 experimental terminal with beam intensity as the optimization objective.Different beams of different energy were used in the experiment,and the results showed that the effects of different optimization methods are quite different: The effect of Bayesian Optimization(BO)method using EI(Expected Improvement)acquisition function is poor,the effect of UCB(Upper Confidence Bound)acquisition function is closely related to its hyper-parameter,and the effect of Differential Evolution(DE)method is not as good as BO method in large search space.Finally,some improvements are made to the DE method,which yielded good results.The work in this thesis has provided technical guarantee for the stable and reliable operation of HIMRF,as well as new reference for the study of automatic beam tuning,and it will provide a good preparation for further online optimization of other beam parameters at HIMRF.