Forecasting energy trends and peak usage at the University of Virginia

Abstract

Forecasting energy trends, especially peak usage, is a valuable and necessary part of energy management. Accurate prediction allows for the control and alleviation of overuse during peak times with the implementation of energy efficiencies. Using hourly kilowatt data from over 200 buildings on the University of Virginia's campus this paper examines the most effective techniques for developing both individual building and overall grid use energy models with a specific focus on predetermining peak usage points. This paper proposes that the expectation-maximization algorithm within the state-space framework is the most effective method for smoothing missing values, a common occurrence when working with metered energy data. Next, this paper covers two separate methods for creating forecasting models. The first, a linear model, was found to be most successful in predicting campuswide energy usage, with the inclusion of features of temperature, humidity, school session, and three temporal variables. The second method, found to be most successful when forecasting short term individual building energy use, uses a seasonal autoregressive integrated moving average (SARIMA) model. Finally this paper delves into the intricacies involved in clustering buildings based on their energy usage trends rather than building use, and the implications that arise from such a process. The conclusions made in this paper can be rescaled and applied to larger energy systems outside the university setting.

Towards sustainable developing cities: A simplified forecasting model for sizing grid-tied PV using monthly electricity bills

“…If the system is too small the probability of the power supply being insufficient increases, however, if the system is too large it results in unnecessary capital expenditure. Between 2006 and 2019 there were several investigations regarding load forecasting through time series analysis (Zala and Abhyankar, 2014;Heylman et al, 2015;Yilmaz et al, 2019;Zhu et al, 2019), and research done regarding optimal energy system design (Yang et al, 2009). These studies and their relevant results are discussed below.…”

“…The papers evaluated various algorithms including regression, clustering, k-means, classification and neural networks, and concluded that a universal protocol capable of evaluating each algorithm for a specific set of input parameters is necessary to determine the optimal technique. Heylman et al (2015) investigated energy usage trends from over 200 buildings on the University of Virginia's campus to determine the most effective techniques for forecasting building energy usage. The study also examines the clustering of buildings based on their energy usage trends rather than the building's functional use.…”

“…Although various load forecasting-and system optimisation techniques have been discussed (Yang et al, 2009;Zala and Abhyankar, 2014;Heylman et al, 2015;Yilmaz et al, 2019;Zhu et al, 2019), none of the solutions present a method for forecasting individual building (or property) load through easily accessible input parameters, essential for the South African educational context, in which smart-meters are not commonly used, access to smart meter data (where used) is difficult to obtain, and the requisite skills for detailed simulations and available resources (time and money) at government level are limited. Furthermore, none of the surveyed forecasting approaches present a method for optimally sizing renewable energy systems from a forecast load profile.…”

Spatio-temporal representation for long-term anticipation of human presence in service robotics

“…Long-term predictions require analysis of time series, which are typically gathered over long time periods. In the time domain, the evolution of a given variable is usually modelled as a combination of trend, seasonal and cyclic patterns [21], [22]. While the trend represents the long-term decrease or increase, cyclic patterns are understood as rises or falls with no periodicity, and seasonality models the periodic changes.…”