基于智能预测控制的Barrier Bucket高频数字低电平系统关键技术研究

移动式脉冲（Moving Barrier Bucket）堆积方案是突破传统束流堆积方式瓶颈，提高重离子流强的有效方法。同步高频系统生成的幅度、相位和周期可调的单正弦或半正弦电压（BB电压）是实现此堆积方式的核心。BB电压的畸变会导致束流的发射度增长、不稳定性、损失及非平滑散束。本项目研究发现BB电压畸变是由BB电压的宽带特性、腔体储能特性、高频系统的非线性、束流负载效应所导致的。在高频系统特性不易改变的情况下，通过研究数字低电平控制算法来实现BB电压校正是解决该问题的唯一途径。BB高频系统的大范围动态多谐波频谱变化是系统设计的难点，采用Hammerstein预测模型解决由高频系统特性及频谱动态变化导致的BB电压畸变；采用多谐波智能自适应迭代学习算法来实现束流负载效应补偿功能。研究有利于拓宽低电平领域的发展方向，结果可用于提高HIRFL-CSR的流强，并为强流装置建设提供理论依据和设计经验。

In order to overcome the limitation of the traditional beam stacking method, the Moving Barrier Bucket (BB) stacking scheme is introduced to accumulate the heavy ion beam to higher intensity. The amplitude, phase and periodic adjustable single or half sinusoidal pulse voltage (BB voltage) produced by the synchrotron Radio Frequency (RF) system are the core of this accumulation mode. The distortion of BB voltage will lead to the beam emittance growth and instability, beam loss and beam non-smooth debunch. In this project, the characteristics of RF system are deeply researched. BB voltage distortion is caused by the broadband characteristics of BB voltage, the energy storage characteristics of the cavity, the nonlinearity of the cavity and amplifier system, and the beam loading effect. In the case that the characteristics of the beam, the cavity and amplifier are not easy to change, the only way to solve this problem is to develop the digital low level RF (LLRF) control algorithm to realize BB voltage correction. Large scale dynamic multi harmonic spectrum variation of BB RF system is a key issue in system design. A LLRF system based on intelligent predictive control algorithm is designed. The predictive control algorithm based on Hammerstein model is used to solve the BB voltage distortion caused by the cavity energy storage, the non linearity of cavity and amplifier. The multi harmonic intelligent adaptive Iterative Learning Control (ILC) algorithm is used to realize the BB RF system beam loading effect compensation function. The LLRF system will be debugged with the RF system, and the control algorithm will be proved and optimized. This research is helpful to broaden the development direction of the LLRF field of accelerators. The research results can be used to improve the beam intensity of HIRFL-CSR. It also provides important theoretical basis and design experience for the construction of high intensity accelerator facility.