With the increasing integration of photovoltaic (PV) systems, challenges in condition monitoring, fault diagnosis, and modeling of PV plants have intensified. To address the cluster division aspect of these challenges, a novel and efficient method using an improved spectral clustering (SC) approach is proposed. The approach starts with preprocessing the PV data to normalize variations in magnitude. The clustering process is then executed using the modified SC algorithm. To validate its effectiveness, comparative analyses are conducted using various clustering indices across different test cases. To comprehensively evaluate the performance of the approach, a comprehensive index is proposed. The results confirm that the method significantly enhances the efficiency and speed of cluster division in large-scale PV plants, making it a promising tool for PV system management.