Modeling and control of transverse coupled bunch mode levels in Indus-2 using artificial neural network

Transverse coupled bunch instability (TCBI) is a major concern at high beam current operations at all synchrotron light sources. Techniques for the mitigation of TCBI include higher order mode tuning of RF cavities, optimization of vacuum chamber designs, increasing the damping rate of beam oscillations, optimization of betatron tune values, and multi-bunch feedback systems. Due to uncertainties, time-variation, and disturbances, the dynamic behavior of accelerators requires further tuning of beam parameters beyond theory-based set points for minimizing the transverse coupled bunch mode (TCBM) instability. In this work, an artificial neural network (ANN) based system is developed to minimize average TCBM levels in the Indus-2 synchrotron light source at the Raja Ramanna Centre for Advanced Technology in Indore, India. The ANN is trained based on various TCBM measurements collected at the Indus-2 for various values of betatron tune and chromaticity in order to learn how to map beam measurements directly to parameters such as optimal betatron tunes and chromaticity values that are sent to a beam feedback control system. The ANN takes as input real-time beam data and is coupled to a feedback controller, thereby creating an adaptive feedback that is able to adjust in real time to variation of the accelerator and beam. We provide a detailed overview of our approach as well as experimental results in which the ANN-guided feedback approach increases the operational beam current of Indus-2 from a limit of ∼170 mA up to ∼230 mA within ∼21 min. We believe that this general method can be useful for a wide range of synchrotron sources operating at high bunch currents.