A machine learning-based method for resolving secondary mirror misalignment in telescope optical systems

Abstract

Abstract Larger aperture and larger field of view represent the evolving demands of astronomical research and the advancing trajectory of telescope technology. For ground-based telescopes, the impact of gravity on larger optical systems becomes increasingly important. For space telescopes and Antarctic telescopes, maintenance challenges are another significant consideration. However, directly resolving the misalignment of optical components can guide the system to make corresponding self-compensation, which is very meaningful for these telescopes. We have focused on resolving misalignments in telescope optical systems over the years. Since 2023, we introduced a new method (RSVA) for evaluating the image quality of optical systems. Although originally conceived to investigate the imaging characteristics of large-field optical systems, our ongoing research has revealed its inherent capacity to encapsulate a plethora of information regarding the optical system’s state. Consequently, we promptly combined it with machine learning to resolve the misalignment of secondary mirror in optical systems, ultimately achieving excellent results. Inspired by this concept, we developed the SVA-type method to resolve misalignments in telescope optical systems. This method achieved a root-mean-square (RMS) misalignment resolution of less than 1 μm & 3.6” for the secondary mirror using only 10 rays. This article will provide a detailed explanation of the methodology and results, as well as explore other interesting possibilities arising from this approach. Drawing upon the SVA theory, we aspire to devise a distinctive approach to resolve misalignment in a telescope’s optical system by leveraging imaging shape.