Generation of typical scenarios for distribution networks in planning stage considering photovoltaic and load growth characteristics

With the increasing integration of distributed rooftop photovoltaic (PV) systems into distribution networks, traditional scenario generation methods based solely on historical PV data have become inadequate. This paper proposes a planning-stage PV scenario generation method to address the challenges of high-penetration rooftop PV integration. The method combines Conditional Generative Adversarial Networks (CGAN) with an improved Bass model to estimate new PV capacity. Load scenarios are constructed by analyzing regional load growth patterns. Typical weather days are classified using Spearman’s rank correlation coefficient to form joint PV-load scenarios, which are then reduced using k-means clustering. The study compares multi-scenario energy storage configuration schemes considering planning-stage scenarios with those based only on historical data predictions. Results demonstrate that the generated planning-stage scenarios align well with future actual operating scenarios. Furthermore, the energy storage configuration scheme considering planning-stage scenarios outperforms the scheme based solely on historical data predictions, indicating the proposed method’s effectiveness in addressing high-penetration PV integration challenges in distribution network planning.