Tree Obstacle Clearing in Power Channels is one of the most important aspects of inspection. Traditional inspections rely heavily on field surveys conducted by personnel, which are highly dependent on operational experience and are inefficient and unsafe, making them unsuitable for the needs of smart grid construction. This paper establishes a real-time tree obstacle analysis method using UAV LiDAR. To meet the real-time requirements of point cloud processing, the UAV LiDAR system is paired with a self-developed onboard computer that interprets point cloud data in real-time during flight. The method involves merging and segmenting point cloud frames, determining target areas using tower records, constructing voxel features for point cloud classification, and utilizing three-dimensional spatial distance analysis for hazard analysis. On-site comparative experiments show that this method can achieve real-time detection of tree obstacles in power channels with high accuracy, significantly improving the quality and efficiency of tree obstacle hazard detection in existing transmission lines and strongly promoting the intelligence of power grid inspections.