Modeling and forecasting uncertainties in power exchange between distributed energy systems and urban grid networks

Integrating distributed energy systems into urban power grids is pivotal for sustainable development. Such a transition toward hybrid power infrastructures, which transforms consumers into prosumers, introduces significant uncertainties in the amount of power the grid may receive from distributed energy sources at any given time due to the intermittency of power generation by distributed energy systems and prosumers’ complex energy usage behaviors. To address this challenge that complicates power generation management, this study proposes a novel systematic approach for quantifying, modeling, and predicting power injections from distributed energy systems into the main grid. The proposed method defines uncertainty as unpredictability rather than variability. It utilizes advanced time series analysis to develop a Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model for characterizing and forecasting uncertainties, using the residuals of a Seasonal Autoregressive Integrated Moving Average (SARIMA) model, which captures the predictable patterns. The method was implemented using historical data on power exchanges between the main power grid and a set of households equipped with photovoltaic (PV) distributed energy systems in Austin, Texas. The results demonstrated that the method effectively and accurately predicted the level of uncertainty in future power exchanges, showcasing its potential for practical application in managing hybrid energy networks.