From Energy Flow Regimes to Regime Dynamics: A Window-Based Hidden Markov Modeling Framework for Photovoltaic–Building Systems

High-resolution measurement data from photovoltaic–building systems enable analyses that extend beyond static energy balances toward the temporal structure of system operation. This study proposes a window-based analytical pipeline to identify operating regimes and their dynamic transitions in an integrated PV–building energy system. The approach combines energy flow decomposition, sliding temporal windows, and hidden Markov modeling to capture short-term behavioral patterns, regime persistence, and transition pathways. The results indicate that system operation is dominated by a limited number of persistent regimes with a pronounced diurnal organization. Regime transitions are structured and directional, typically occurring through intermediate operating states rather than via abrupt switching between energetically extreme conditions. The analysis of regime residence times reveals distinct characteristic temporal scales associated with different operating modes. Overall, the proposed framework provides a process-oriented representation of PV–building system dynamics that enhances interpretability and supports advanced monitoring and analysis of integrated energy systems.