Two-stage uncertainty modeling for temporal energy scheduling during extreme weather conditions

The inherent intermittency of distributed renewable energy sources (DRESs) introduces significant uncertainties, necessitating robust uncertainty modeling for the resilient operation of energy networks. As DRESs are weather-dependent, extreme weather events, such as storms, affect real-time operation of energy networks, leading to more power outages. This paper introduces a novel time-coordinated strategy (TCS) using a Bayesian network that dynamically captures temporal correlations among uncertain parameters. The proposed method is compared with traditional Markov Chain Monte Carlo techniques, showing a notable shift in the probability density function under adverse weather events. Furthermore, an intelligent two-stage stochastic energy scheduling mechanism is proposed, which balances the computational efficiency with the scheduling accuracy by intelligently triggering hourly and quarterly energy dispatches for the efficient operation and control of the energy system. Applying our approach to a 33-bus distribution network (DN) shows that TCS maintains a minimal voltage deviation (within 10% of the nominal voltage) during extreme weather, increases the energy sold to the grid, and accurately models weather parameters for better energy scheduling. Further, we introduce two factors, the renewable reliability factor (ℜ) and carbon emission factor (CEF), to validate the performance of renewable energy DNs. A value of ℜ closer to 1 (i.e., 0.976) and that of CEF closer to 0 (i.e., 0.0136) indicates that the proposed approach ensures real-time robustness, reduced carbon emissions, and improved voltage profile even during extreme weather conditions.