With the large-scale integration of distributed photovoltaics, the morphology and characteristics of distribution networks are becoming increasingly complex. Accurate and efficient modeling of distributed photovoltaic aggregation is of significant importance for the planning and operation of new distribution networks. In this study, we first evaluated the similarity of different photovoltaic generation units by integrating the output characteristics of photovoltaic systems and the sensitivity indicators of grid connection points, proposing a distributed photovoltaic clustering method based on the K-means algorithm. Secondly, the output of the photovoltaic aggregation cluster was calculated using the capacity-weighting method. An equivalent network connecting various photovoltaic clusters was constructed based on the node elimination of the admittance matrix, and a modeling method for distributed photovoltaic aggregation was developed. Finally, the effectiveness of the method proposed in this paper was verified through the IEEE-33 test system. The results show that the distributed photovoltaics aggregation modeling method proposed in this paper can effectively simplify network topologies and improve the accuracy of the calculation results while also improving the efficiency of power flow analysis.

Accurate model aggregation is pivotal for the efficient dispatch and control of massive distributed photovoltaic (PV) and energy storage (ES) resources. However, the lack of unified standards across equipment manufacturers results in inconsistent data formats and resolutions. Furthermore, external disturbances like noise and packet loss exacerbate the problem. The resulting data are massive, multi-source, and heterogeneous, which poses severe challenges to building effective aggregation models. To address these issues, this paper proposes a hierarchical aggregation method based on multi-source multi-scale data fusion. First, a Multi-source Multi-scale Decision Table (Ms-MsDT) model is constructed to establish a unified framework for the flexible storage and representation of heterogeneous PV-ES data. Subsequently, a two-stage fusion framework is developed, combining Information Gain (IG) for global coarse screening and Scale-based Trees (SbT) for local fine-grained selection. This approach achieves adaptive scale optimization, effectively balancing data volume reduction with high-fidelity feature preservation. Finally, a hierarchical aggregation mechanism is introduced, employing the Analytic Hierarchy Process (AHP) and a weight-guided improved K-Means algorithm to perform targeted clustering tailored to the specific control requirements of different voltage levels. Validation on an IEEE-33 node system demonstrates that the proposed method significantly improves data approximation precision and clustering compactness compared to conventional approaches.