Control and assembly of complex bend magnet for proposed NSLS-II upgrade

The NSLS-II upgrade program investigates the imple-mentation of complex bend magnets based on permanent magnet quadrupoles (PMQs) to achieve ultra-low emit-tance and enhanced brightness. While PMQs provide high field gradients and compact lattice configurations, they also introduce challenges in tunability, thermal stability, radiation resistance, and field quality control. This paper presents progress on the design and assembly of Halbach-style magnets constructed from 16 permanent magnet (PM) wedges. The control system is designed to achieve precise magnetic field quality through a nonlinear multi-input multi-output (MIMO) optimization framework. To address this nonlinear MIMO challenge, machine learning approaches are proposed to support assembly objectives. To enable ML inference, edge AI hardware platforms are evaluated and selected based on the specific requirements of the control system.