Photovoltaic Panel Modeling is one of the key technologies for monitoring the state and evaluating the performance of photovoltaic power systems. In the process of 3D LiDAR point cloud modeling, point cloud registration plays a crucial role. To address the limitations of the traditional Iterative Closest Point (ICP) algorithm, such as high requirements for initial position, slow computation speed, and lack of robustness in point-to-point correspondence search, an optimized method is proposed by combining the Four-Point Congruent Sets (4PCS) algorithm and the Generalized Iterative Closest Point (GICP) algorithm. This approach incorporates three-dimensional voxel grid downsampling and bidirectional K-Dimensional tree (KD-tree) for accelerated nearest neighbor point search, enabling efficient and accurate registration of photovoltaic panel point clouds. Experimental evaluations using simulated data and actual photovoltaic panel scans with LiDAR demonstrate significant improvements in registration accuracy and computational efficiency compared to the traditional ICP algorithm.