Robust Optimization of Renewable Energy Systems Considering Interannual Variability : Evaluation of Life-Cycle Cost and Self-Sufficiency Reliability Using Stochastic Demand Scenarios

This study evaluates interannual variability (IAV) in robust sizing of grid-connected PV–battery systems for school buildings. A Ridge-kNN model generates stochastic demand scenarios from one year of measured data and 20 years of weather data. Under rule-based control, NSGA-II optimizes PV area, battery capacity, and inverter capacity by LCC, SSR, and achievement reliability. Results show that high SSR sharply increases battery capacity and LCC, while single-year optimization can underestimate long-term achievement risk.