Abstract: In the context of large-scale photovoltaic grid integration, traditional reactive power compensation technologies predominantly rely on fixed rules or empirical methods, which install fixed-capacity compensation devices at specific locations. However, these approaches struggle to accommodate the randomness and volatility of photovoltaic power output, resulting in suboptimal compensation performance. To address this, this study investigates reactive power compensation technology for 10 kV distribution network substations under photovoltaic grid integration. Based on the topology structure, load characteristics, and photovoltaic connection points of the distribution network, key reactive power compensation nodes are identified. Using these identified nodes as a foundation, the study calculates the reactive power compensation capacity for substation areas by considering dynamic variations in photovoltaic generation and load during different time periods. Suitable compensation devices are then selected according to the identified nodes and calculated capacities to design centralized compensation solutions. Application analysis results demonstrate that implementing this technology achieves an average voltage deviation of only 0.16% across six nodes, effectively controlling voltage deviation and significantly improving voltage quality. Additionally, the daily average power factor in substations improves markedly to the 0.96~0.98 range, effectively reducing line losses while delivering substantial economic and social benefits.