Abstract: To address the challenges of insufficient safety belt monitoring accuracy and frequent falls caused by dual-hook interlock failures in high-altitude power transmission projects, this study proposes an intelligent analysis technology integrating RTK centimeter-level positioning with multi-camera visual perception. By integrating an RTK module and binocular cameras into smart safety helmets, the system achieves dynamic positioning accuracy of ≤5 cm. Combined with cross-camera collaborative tracking through ground visual networks, it overcomes blind spots in obstructed environments. The end-edge-cloud collaborative early warning system enables millisecond-level response to dual-hook interlock failures. Verification at a 500 kV substation demonstrated: 95.3% accuracy in hanging point recognition, reduced early warning response time to under 200 ms, and a 67% decrease in high-altitude fall incidents.