机器学习在储存环轨道校正中的应用研究

X射线同步辐射光源，是现代科学研究中最强大的工具之一。位于中国上海的上海光源，是一台能量为3.5 GeV的先进的第三代中能同步辐射光源。第三代同步辐射光源要提供高亮度、高稳定性的同步辐射来满足实验条件要求苛刻的前沿研究，因此对束流的轨道稳定性有很高的要求。为此，采用机器学习算法进行电子束轨道的控制和反馈。这种基于神经网络的轨道校正方法不依赖于具体的响应矩阵，建立非线性映射关系，并且还可以进行连续的在线再训练，对上海光源的轨道校正和提高束流轨道稳定性有重要意义。

Application of machine learning in orbital correction of storage ring

Synchrotron light source is one of the most powerful tools in modern science and technology. Shanghai Synchrotron Radiation Facility (SSRF), located in Shanghai, China, is an advanced 3.5 GeV 3rd-generation medium energy light source. The 3rd-generation synchrotron radiation light source will provide high brilliance and high stability synchrotron radiation to fulfill the advanced experimental conditions in frontier researches. To achieve highly stable radiation, it is important to have highly stable beam orbit. Thus we adopted machine learning method to control and feedback the orbit. Using this neural network-based orbit correction method, which doesn’t rely on the response matrix, we can establish a nonlinear mapping relationship between correctors and the orbit distortions and perform continuous online retraining. This new method can significantly improve the orbit stability of SSRF.