Performance Evaluation of Forecasting Strategies for Electricity Consumption in Buildings

Abstract

In this paper, three main approaches (univariate, multivariate and multistep) for electricity consumption forecasting have been investigated. In fact, three major algorithms (XGBOOST, LSTM and SARIMA) have been evaluated in each approach with the main aim to figure out which one performs the best in forecasting electricity consumption. The motivation behind this work is to assess the forecasting accuracy and the computational time/complexity for an embedded forecasting and model training at the smart meter level. Moreover, we investigate the deployment of the most efficient model in our platform for an online electricity consumption forecasting. This solution will serve for deploying predictive control solutions for efficient energy management in buildings. As a proof of concept, an already existing public dataset has been used. These data were mainly collected thanks to the usage of already deployed sensors. These provide accurate data related to occupancy (e.g., presence) as well as contextual data (e.g., disaggregated electricity consumption of equipment). Experiments have been conducted and the results showed the effectiveness of these algorithms, used in each approach, for short-term electricity consumption forecasting. This has been proved by performance evaluation and error calculations. The obtained results mainly shed light on the challenging trade-off between embedded forecasting model training and processing for being deployed in smart meters for electricity consumption forecasting.

Multi-Step Hourly Power Consumption Forecasting in a Healthcare Building with Recurrent Neural Networks and Empirical Mode Decomposition

“…Thus, the results achieved in this work may only be compared with a few studies. In [ 10 ], hourly consumption of a household was predicted with a multivariate model. The best result was provided by a seasonal ARIMA: an MAPE of 1.162%.…”

“…Different models should be developed for different customers with different consumption patterns and energy needs. Thus, besides research devoted to predicting overall hourly or daily demand [ 1 , 2 , 3 , 4 , 5 ], other studies have focused on specific consumers [ 6 , 7 , 8 , 9 , 10 ], and different forecasting tools have been proposed to obtain reliable demand predictions. These predictions may have a one-step-ahead horizon or a multi-step one.…”

“…These predictions may have a one-step-ahead horizon or a multi-step one. Despite that this last structure seems the most advisable option, there are few researchers studying it [ 1 , 8 , 10 , 11 , 12 ], a fact that is hardly surprising considering that it represents a more challenging problem. That is why the models studied in this work will provide hourly predictions of the electric energy that a specific consumer will demand in the following 24 h, i.e., 24 predictions will be provided all at once.…”

“…As pointed out above, short-term forecasting of electric energy demand has become a key issue for companies purchasing electricity because of the rising and fluctuation trend of prices, which forces them to continuously adjust their purchasing bids to the fast-changing evolution of those prices; not only is the short-term forecasting of the overall energy demand necessary, but also a more specific one focused on the electric load of specific companies, institutions or buildings is needed, as those medium-small consumers also demand an accurate prediction of their energy needs to optimize their electricity purchases. Several works have been devoted to providing reliable forecasting tools for those customers [ 6 , 7 , 8 , 9 , 10 ]. Nevertheless, there are not many papers devoted to forecast consumption of those specific consumers because data needed to carry out those tasks must be actual measurements provided by specific equipment and belong to consumers who are not always willing to share them with time series researchers.…”

Energy Consumption Forecasting for Smart Meters Using Extreme Learning Machine Ensemble

“…Smart meters not only enable occupants to have insights of their own consumption patterns, but also provide useful information to energy suppliers in order to perform better planning of energy load. In this scenario, energy forecasting is considered an important tool for planning and decision making processes [ 6 ]. Its main challenge, however, is the high volatility of data concerning individual households.…”