Probabilistic Load Forecasting Based on Adaptive Online Learning Article Swipe
YOU?
·
· 2021
· Open Access
·
· DOI: https://doi.org/10.1109/tpwrs.2021.3050837
· OA: W3106773247
Load forecasting is crucial for multiple energy management tasks such as\nscheduling generation capacity, planning supply and demand, and minimizing\nenergy trade costs. Such relevance has increased even more in recent years due\nto the integration of renewable energies, electric cars, and microgrids.\nConventional load forecasting techniques obtain single-value load forecasts by\nexploiting consumption patterns of past load demand. However, such techniques\ncannot assess intrinsic uncertainties in load demand, and cannot capture\ndynamic changes in consumption patterns. To address these problems, this paper\npresents a method for probabilistic load forecasting based on the adaptive\nonline learning of hidden Markov models. We propose learning and forecasting\ntechniques with theoretical guarantees, and experimentally assess their\nperformance in multiple scenarios. In particular, we develop adaptive online\nlearning techniques that update model parameters recursively, and sequential\nprediction techniques that obtain probabilistic forecasts using the most recent\nparameters. The performance of the method is evaluated using multiple datasets\ncorresponding with regions that have different sizes and display assorted\ntime-varying consumption patterns. The results show that the proposed method\ncan significantly improve the performance of existing techniques for a wide\nrange of scenarios.\n