Design of a dual-mode transverse deflecting structure using neural network and multiobjective algorithms

Abstract

Shanghai Synchrotron Radiation Facility/Shanghai Soft X-ray FEL Facility is currently developing an advanced variable polarization transverse deflecting structure TTDS (two-mode transverse deflecting structure) using a dual-mode rf structure concept. Driven by two different rf power sources, this novel TDS works using both the ${\mathrm{HEM}}_{11}$ and ${\mathrm{HEM}}_{12}$ modes for simultaneous vertical and horizontal deflections, consequently, it can provide a time-varying polarization of the electrical field at ultrafast speeds. It is capable of producing circular and elliptical polarizations as well as flexible vector combinations through amplitude and phase modulation from a low-level rf system. The work presented in this paper is focused on the analysis and design of the variable polarization TDS, consisting of dual-mode cells and two dual-mode couplers. Designing and optimizing for dual-mode design and optimization is complex; consequently, an advanced optimization procedure based on neural networks and multiobjective algorithms has been developed. This improves the accuracy and efficiency of the rf structure design process. Through iterations, the dual-mode cells in the final design are optimized for high impedance and other rf performance criteria for both the ${\mathrm{HEM}}_{11}$ and ${\mathrm{HEM}}_{12}$ modes. The two couplers for rf power input and output are also optimized. Based on the optimized design and rf sensitivity analysis, the mechanical design has been completed and is now ready for manufacture.