Generation Method of Typical Light-Load Correlation Scenario Set for Photovoltaic Hosting Capacity Assessment

Abstract

At present, photovoltaic power generation has been widely used and more photovoltaic power stations are constantly connected to the power system, which put forward higher comprehensive requirements for the planning and operation of power system on the macro scale. From the point of view of data feature mining and correlation, this paper proposes a method for generating typical scene sets of photovoltaic load correlation taking into account meteorological factors. Firstly, HDBSCAN clustering is used to cluster the comprehensive operation data of the system in the planning period and to screen out typical scenarios. Then different typical scenarios are fused by FP-growth correlation using meteorological characteristics. Finally, a typical set of associated scenarios including the operation of photovoltaic power system is obtained. Taking a city power grid as an example, the correlation typical scene method and the traditional whole-year time sequence method are compared and analyzed. The scenario set of this method contains a series of load and photovoltaic indicators to provide a more comprehensive macro description of the scenario. The experimental results show that the scene set generated by this method can improve the computational efficiency while maintaining a good accuracy. This provides more references for planners on medium and long-term time scale for PV-based power system data, improves the comprehensiveness and scientificalness of planning, and meets the needs of power system for medium and long-term data scenarios.