Systematic investigation and improvement of LLRF controls for LINACs and cyclotrons

点击翻译按钮获取中文摘要

This proposal requests funds to support a program in research and development in accelerator science. The program is aimed at developing innovative solutions to address three major challenges in the control of RF cavities as follows: 1.Microphonics suppression is crucial to improve the beam quality, which is considered as a major challenge by many accelerators operating in continuous wave (CW) mode. Accelerators have achieved a suppression of a factor of 2-8 by phase error feedback to an installed fast acting piezo element. To improve the suppression further, how the different mechanical modes affect the beam must be understood. An RF analysis evaluating the different modes will provide the information regarding the modes that need to be actively suppressed. An analysis of the mechanical structure of the system further provides the necessary information regarding observability and controllability of each mode. The latter two points provide the limitations of active microphonics control with a piezo element, which are unknown to the accelerator community. An evaluation of a possible control system based on a collocated sensor/actuator can improve the accuracy. 2.TRIUMF's e-LINAC will be operated in the CW mode. The acceleration cavities are operated in vector sum mode. While CW operation and vector sum control are operational concepts which can be found in accelerators, the combination of the two is rather new and accompanied by new challenges. Lorentz force induced parametric oscillations can be a limiting factor for the operational gradient of RF cavities. An analysis of Lorentz force induced oscillations in coupled cavities is necessary to understand the stability of such systems. An active oscillation cancellation system that relies on a collocated sensor/actuator pair will be developed to increase the system stability, which may be applicable to other accelerator laboratories worldwide. 3.Sparks in cyclotrons occur on a regular basis. They have to be detected and reacted upon for maximum machine protection and minimal down time. Machine learning is gaining momentum in the field of particle accelerators as the different operational areas for such systems and the benefits are evaluated and learned. Fault diagnostics, can be especially important for large and complex machines. We are proposing to investigate a spark detection system based on machine learning. The first two points address the field stability of an RF superconducting cavity and hence improving the beam stability, the third point addresses artificial intelligence towards accelerator LLRF control. With experience gained it can be applied to the above project parts in terms of more intelligent data collection and control. The proposed project is a perfect training environment for young researchers in the fields of science and engineering. The proposal requests funds to support graduate students' salaries and funds to support travel to conferences for young researchers.