Vision-Guided Intelligent Robotic System for In-core Thermocouple Maintenance in Nuclear Reactors

Abstract The Maintenance of reactor core thermocouples, especially online welding operations following component replacement, poses critical challenges, including its stringent precision requirements, specialized welding processes, complex implementation conditions, and limited workspace. In response to these challenges, this paper proposes an online intelligent welding robot system. The system is primarily composed of 5 parts: a three-stage high-precision visual perception system, a six-axis robotic arm with adaptive tracking, a dedicated laser welding system, a control and interactive center, and auxiliary equipment. The research achieves an overall positioning error of ≤0.1 mm at a variable distant and an operational time within 3 minutes, which can ensure automated, standardized, and remotely controlled high-precision welding repair. Preliminary validation on a dedicated simulator platform on 1:1 scale to the real scene confirms its effectiveness, marking it as the first complete system in the domestic nuclear power sector for automated repair of reactor core thermocouples. This technology introduces a new method for the intelligent operation and maintenance of critical nuclear power plant equipment and can be extended to other high-precision repair applications.