Gas-insulated Switchgear (GIS) is a critical core component of power systems. In China, 1000 kV ultra-high voltage transmission projects exclusively utilize GIS, and its operational reliability is directly tied to the power supply security of the system. The fully enclosed structural characteristics of ultra-high voltage GIS necessitate the use of live detection methods, including ultrasound, ultra-high frequency (UHF), and X-rays. Among these, ultrasound and UHF partial discharge detection, combined with time-difference localization, can determine the approximate location of partial discharge sources but cannot accurately identify the type of discharge. Multi-angle X-ray imaging, supported by specialized image processing and recognition technologies, enables the "visualization" of internal GIS structures and rapid diagnosis of defect conditions without power interruption or disassembly. This significantly improves the accuracy of fault localization and defect identification in GIS equipment. However, previous studies have primarily relied on scaled-down GIS models at lower voltage levels and standard metal particle defects. This paper establishes an experimental platform for full-scale ultra-high voltage GIS, utilizing metal particles generated by real-world cutting and knob operations to investigate the impact of X-rays on UHF partial discharge caused by metal particle defects in ultra-high voltage GIS.