Long-Term Regional Hourly Net Load Profile Generation with Behind-the-Meter PV

Long-term load forecasting is essential for the planning and operation simulation of power grid. However, most existing approaches either can only predict very general metrics, such as maximum load, or are purely data-driven and unable to adapt to changes in load magnitude and load patterns. This work proposes a long-term load profile generation framework that takes distributed photovoltaic (DPV) into consideration. The net load is decomposed into actual load and DPV output, each modeled separately. The experimental results show that our proposed method better captures the shifting load patterns, and a detailed analysis is conducted on the generated load profile of Guangzhou in year 2030 under various DPV capacity scenarios.