A Novel Clustering Method for Extracting Representative Photovoltaic Scenarios with a Focus on Peak Moments

Against the backdrop of energy shortages and global warming, photovoltaic (PV) power generation, as a major form of renewable energy, helps reduce carbon emissions but also introduces significant uncertainty into the power system. Therefore, extracting representative PV scenarios is crucial for improving optimization efficiency. To tackle this problem, we consider using the attribute importance and density weight of data points to measure the peak moments of PV scenarios, thereby effectively extracting representative PV scenarios. We also introduce an alternating optimization algorithm based on Lagrange multipliers and eigenvalue decomposition for solving this problem. Using real output data from PV systems in Hebei, we validate the proposed clustering model’s high cohesion, low coupling, and its effectiveness in preserving the peak moments of the original PV scenarios.