Photovoltaic Power Load Forecasting Algorithm based on Data Fusion Cluster Tree

With the rapid development of new energy technology, photovoltaic load forecasting is one of the current research emphases and hot spots. Optical image processing is used to study the regional situation in photovoltaic load forecasting. The primary objective of point cloud alignment is to determine the optimal spatial correspondence among two or more point cloud datasets. Despite advancements in this research area, a persistent issue remains: the features extracted by point cloud alignment algorithms are often insufficiently rich, hindering accurate alignment. This study aims to effectively exploit information from different resolution levels in point cloud data, enhancing alignment accuracy. Thus, we propose a multi-scale feature fusion point cloud alignment algorithm (MSM, Multi Scale Matcher) that incorporates the cross-attention mechanism. Experiments conducted on the ModelNet40 dataset demonstrate that the proposed method significantly enhances the accuracy of load forecasting.