We consider an increasingly popular demand-response scenario where large-load customers, e.g., datacenters, utilize energy storage to reduce the peak procurement from the grid, which accounts for up to 80% of their electric bills. We focus on minimizing the peak-demand charge using energy storage under the online setting, where the loads and renewable generations are revealed sequentially in time but we have to make irrevocable decisions at current epoch with little or no future information. Such an online problem is uniquely challenging due to (i) the coupling of irrevocable decisions across time imposed by the inventory constraints and (ii) the noncumulative nature of the peak procurement. We tackle this issue by developing an optimal online algorithm for the problem that attains the best possible competitive ratio (CR) among all deterministic and randomized algorithms. We show that the optimal CR can be computed in polynomial time, by solving a linear number of linear-fractional problems. More importantly, we generalize our approach to develop an anytime-optimal online algorithm that achieves the best possible CR at any epoch, given the inputs and online decisions so far. The algorithm retains the optimal worst-case performance and achieves adaptive average-case performance. Simulation results based on real-world traces show that, under typical settings, our algorithms improve peak reduction by over 19% as compared to baseline alternatives.
This is a joint work with Yanfang Mo, Qiulin Lin, and Joe Qin, all from City University of Hong Kong.
Bio:
Minghua Chen is a Professor at School of Data Science, City University of Hong Kong. His research interest is in online optimization, machine learning, energy systems, transportation, and networked systems.
https://www.mhchen.com
