The growing emphasis on power quality has posed significant challenges for distribution system operators (DSOs). Among these challenges, short-term voltage fluctuations, specifically voltage sag, have drawn considerable attention. In this study, three concepts of average edge (AE), lower average edge (LAE), and upper average edge (UAE) based on the electrical connection matrix and voltage magnitude sensitivity matrix are defined and used as the partitioning first level. At the second level, a kernel smoothing function is employed to refine the zoning process. Subsequently, strategic locations within each zone are identified: the vertex and middle buses. These carefully selected buses serve as installation points for dynamic voltage restorers (DVRs). In response, this study proposes a novel solution by partitioning the distribution network into distinct zones. The focus lies in developing a two-level offline partitioning approach for active distribution networks (ADNs) that incorporate photovoltaic (PV) systems. To evaluate the effectiveness of the proposed method, numerical studies were conducted on modified IEEE 33-bus, IEEE 69-bus, and Iranian 95-bus systems, with simulations performed usingMATLAB/Simulink. The proposed method provides good performance and fast calculation speed for distribution network partitioning, as confirmed by the results. Test results show improved bus voltage with PV unit integration. Additionally, power loss in the IEEE 33-bus, IEEE 69-bus, and Iranian 95-bus networks decreased by 47.73 kW, 56.87 kW, and 69.63 kW, respectively. Furthermore, the voltage profile improved from 0.75 p.u. to 0.928 p.u. during a voltage sag in the IEEE 33-bus system, and in steady state, the voltage increased from 0.933 to 0.959 p.u.
