Di Capua, Vincenzo (2022) Real-time magnetic measurement system for synchrotrons control. [Tesi di dottorato]

Preview

Text vincenzo_dicapua_34.pdf Download (28MB) | Preview

Collection description

The thesis focuses on the design, implementation and validation of a distributed real-time magnetic measurement system tailored for particle accelerators’ magnets. In particular, the thesis focuses on the study of 4 macro-areas connected to this system. The first study regards the development and the implementation of a real-time measurement system starting from the system requirements defined by the final users (i.e. low level radio frequency, power converters and machine operators). To satisfy all the necessities and all the constraints for all the seven particle accelerators that will benefit from this system, it was a crucial point have a measurement system as much flexible as possible to minimize the maintenance costs and to being prone to future requests by the system’s users without the necessity of massive changes. In the thesis, a novel system, one of a kind, able to measure and simulate a magnetic field in real-time and provide it over a optic-fiber based network is presented. To satisfy all the requests and to overcome all the constraints it was necessary to design new custom electronics, both new PCBs modules to be integrated in the framework of the standard CERN electronics, and new FPGA modules described in VHDL. Moreover it was necessary to design new software modules integrated in the software Framework to be compatible with all the CERN infrastructure. The second study concerns the development of a necessary tool to monitor in real- time the magnetic field provided by the seven installed systems over the optic-fiber. This tool was necessary for three reasons: first, to characterize and calibrate the system looking at the same output that the users will receive. Second, to monitor the correct behavior of the systems. Third to speed-up the debugging in case of issues. In this layout a novel tool based both on commercial National Instruments and custom hardware is proposed. The tool was connected to all the seven systems thanks to an optic multiplexer. The third study regards the DC and dynamic performance evaluation of the pre- sented system. A satisfying agreement with the metrological requirements for the system was found after the fine calibration of the systems. The fourth study concerns the possibility to use neural networks to predict the magnetic field and all its non linearity such as eddy currents and hysteresis inside magnets for particle accelerators. Considering a calibration quadrupole as a case study various neural network based architectures were designed and implemented. The achieved result of this study is a network able to predict the magnetic field leading to a percent error below 0.02%. A detailed study of these four topics is presented. All the realized systems and subsystems were benchmarked with simulation and experimental measures performed all around the CERN acceleration complex (i.e.PSB,PS,SPS,AD,ELENA and LEIR). A satisfying agreement with respect to the original system’s requirements was found in all cases.

Downloads

Actions (login required)

View Item