Session: 08-01-01: Environmental Impact of Energy Systems, Components Optimization, and AI for Energy Systems

Paper Number: 145349

145349 - Short-Term Time Series Energy Forecasting for Smart Buildings Using Machine Learning Models 

The energy consumption by the building sector is rapidly increasing and is projected to reach 50% of global energy production by 2050, raising concerns about escalating carbon emissions. Consequently, there has been a surge in energy-efficient smart buildings aiming to optimize energy usage and reduce carbon footprints. The ‘smart’ factor is achieved by deploying Internet of Things (IoT) devices that constantly monitor the building’s energy consumption, offering increased control over the energy resources. Furthermore, the data from these devices can also be used to forecast future energy demands to guarantee sufficient energy production levels. In recent years, extensive research has been conducted to achieve accurate forecasting techniques through Machine Learning (ML) models. Short-term energy forecast models that predict energy demand within 1 hour to 1 week ahead have gained popularity due to their effectiveness in residential buildings. These models help building managers anticipate demand fluctuations, enabling them to allocate energy more efficiently, prevent energy waste, and adapt energy production strategies accordingly.

Additionally, better forecasting allows the integration of renewable energy sources like solar and wind, ensuring a more sustainable energy mix. This paper compares short-term time series energy forecasts for smart buildings using multiple machine learning models. We explore various statistical, decision-tree-based, and gradient-boosting methods. The data is collected from two energy meters deployed within an IoT laboratory measuring the energy consumption of the laboratory’s constituents – such as the workstations, four HVAC systems, a pantry area, one split-AC, to name a few- at a 15-minute interval. The energy meters can measure up to 6 three-phase circuits and 18 single-phase circuits. A weather station also provides temperature, humidity, and air quality data. These datasets are combined, processed, and analyzed for their temporal behavior. The data is then inputted into the ML models for training and testing, thereby forecasting the energy demands for the next 15-minute to 1-week time horizon. Each ML model is hyperparameter-tuned and cross-validated to fit the nature of the dataset better. The performance of the ML models for the different time horizons is evaluated using different performance metrics such as R2 score, RMSE, and MAPE. Accurate energy forecasting will implement better energy management systems through reduced production and operational costs, better power output scheduling, and improved user comfort. Through these advancements, the building sector can contribute significantly to achieving global energy efficiency goals. Ultimately, this approach leads to more sustainable infrastructure and a reduced environmental impact, benefiting both businesses and society.

Presenting Author: Jayakumar Vandavasi Research and Development Centre, Dubai Electricity and Water Authority

Presenting Author Biography: Eng. Jayakumar Vandavsi Karunamurthy ( Jayakumar) has been serving as the Senior Principal Researcher for 4IR Area Lead Since 2021 and he earned master degree engineering in applied electronics at Anna university ,India.

Eng. Jayakumar has more than 28 years of experience in various fields like Medical R&D, Sub- Sea ROVs, and as an IoT Solution Architect. he has leveraged his extensive experience and skill set to drive meaningful progress in DEWA R&D journey towards Industry 4.0 readiness to achieve the DEWA 2030 Net zero Goal.

Authors:

Rufaidah Chikte Research and Development Centre, Dubai Electricity and Water Authority
Ansu Mathew Research and Development Centre, Dubai Electricity and Water Authority
Balaji Ramachandran Research and Development Centre, Dubai Electricity and Water Authority
Khuloud Almaeeni Research and Development Centre, Dubai Electricity and Water Authority
Nawal Aljasmi Research and Development Centre, Dubai Electricity and Water Authority
Jayakumar Vandavasi Research and Development Centre, Dubai Electricity and Water Authority