The current traditional grid inspection method eliminates potential hidden problems of the power system by ranking fault points, but the lack of extraction and fusion of grid scene features leads to poor inspection results. In this regard, a three-dimensional model fusion-based inspection method for an extra-high voltage power grid is proposed. The key motion metric function is defined by calculating the local minima. And the 3D model is constructed to extract and fuse the UHV grid scene feature data, and finally, the path of UAV inspection is planned. In the experiment, the inspection performance of the proposed method is verified. The analysis of the experimental results shows that when the proposed method is used to inspect the extra-high voltage grid, the KRS index value is high and has a more excellent UAV inspection path planning capability.