Description
Advanced forecasting methods properly coupled with optimization techniques could play a pivotal role toward improving modern energy management systems (EMS) to make superior informed decisions. Scenario generation-based forecasting and stochastic optimization are useful tools for improving buildings and microgrid’s EMS performance through incorporation of uncertainties in energy resources,demand, and supply. These methods have shown promise particularly in problems requiring collaborative agent involvement and high degrees of uncertainty. The degree of impact that scenarios can have on decision-making benefits is believed to be strongly associated with the quality of the scenarios. Throughout the literature on probabilistic forecasting, classical verification methods, such as the well-known continuous ranked probability score (CRPS), are widely used. However, there is little study on the evaluating the performance of scenarios sets in a downstream task, i.e., the real-time functionality of EMSs. The present study proposes an approach leveraging the prior information about the EMS performance and validates the scenario generation methods in an online environment for a multi-agent battery scheduling
case study. It considers discrepancy detected between theoretical and practical performances of scenario sets and uses evaluation methods in the context of downstream utility in terms of energy cost, carbon emissions and grid stability. The proposed integrated approach can significantly enhance the capabilities of stochastic
optimization in energy management applications and offer helpful information for energy management system designers and decision-makers.
Period | 11 Oct 2023 |
---|---|
Event title | Global Cleaner Production Conference |
Event type | Conference |
Location | Shanghai, ChinaShow on map |
Degree of Recognition | International |