Abstract: Regular rust removal and corrosion protection for the outer walls of large oil and gas storage tanks is critical to their safe operation. Conventional manual maintenance methods are characterized by low efficiency, high safety risks, and limited adaptability to irregular structures such as stiffening rings and wind girders. To address these challenges, this paper develops a magnetic adsorption wall-climbing robot system that integrates autonomous navigation, multimodal operation and intelligent control. The system is of a modular design, including a high-strength magnetic adsorption mobile platform, a six-degree-of-freedom collaborative robotic arm, a constant-force flexible rust-removal end-effector, and a recyclable abrasive blasting unit. A multi-sensor navigation system combining Simultaneous Localization and Mapping (SLAM), vision sensors, and an Inertial Measurement Unit (IMU) enables high-precision positioning and adaptive path planning in complex tank-wall environments. In addition, dedicated obstacle-crossing tools and adaptive operation algorithms are designed for irregular structures. Experiments show that the proposed system achieves a rust removal efficiency of 20 ~ 30 m²/h while satisfying the Sa2.5 surface preparation standard. It can reliably cross obstacles and effectively process irregular structures, significantly improving the automation level and operation safety of storage tank maintenance.