利用改进型共轭方向搜索法在线自动优化粒子加速器性能的应用研究

在粒子加速器机器性能优化中，通常采用手动线性扫描方式优化相关重要物理参数。该方法通常有效，但费时费力，并且随变量增多其有效性大大降低。而随着计算机技术及优化算法的不断发展，建立强有力的多维变量空间在线自动优化手段和工具成为可能，且十分必要。改进型共轭方向搜索法（RCDS）是最近由X. Huang提出的一种在线优化算法，具有对噪声容忍度高、收敛较快等优点。本项目计划基于北京正负电子对撞机重大升级改造工程（BEPCII），开展利用RCDS方法在线自动优化加速器性能的应用研究。在BEPCII上争取实现日常例行的亮度及其他重要物理参数的扫描优化机制，进一步提升整体机器性能。RCDS方法有广泛的适用性，通过掌握RCDS方法的具体实现，积累实践经验，未来可以将该方法扩展至国内其他加速器重要物理参数的在线优化研究，对提升我国加速器性能，实现预期设计指标，具有十分重要的实际意义。

In the optimization of the machine performance of particle accelerators, manual and linear tuning approach is usually used, which can be effective in many cases, but is time-consuming, and whose effectiveness decreases as the number of variables increases. On the other hand, along with the continuous development in computing technique and in optimization algorithms, to build an effecitve, online automatic optimization, is feasible and necessary. One online optimization method, so-called robust conjugate direction search (RCDS) method, was proposed recently by X. Huang, which has the advantages of high tolerance to noise and rapid convergence. In this project, the applicants will apply the RCDS method to the online automatic optimization of the luminosity and other important parameters of BEPCII. Efforts will be made to realize routine RCDS scan and optimization of the luminosity and other parameters, to further improve the machine performance of BEPCII. RCDS method has wide applicability. Through this study we will master the concrete implementation of the RCDS method and accumulate practice experience, which will benefit the applications of this method to the online optimization of the important parameters of other accelerators in China.