This dissertation was written as a part of the MSc in Data Science at the International
Hellenic University.
The recent exponential growth of available data in today’s fast-paced technology-driven
world has made systematic data collection and the concurrent extraction of meaningful
information through them a necessity. In light of the adverse effects of climate change,
data analysis and AI techniques can be directly applied to provide forecasts that could be
used for the efficient scheduling and operation of energy usage. Especially in the built
environment, Energy Load Forecasting (ELF) enables Distribution System Operators
(DSO) or Aggregators to accurately predict the energy demand and generation tradeoffs.
Today’s near-zero Energy buildings (nZEBs), which are buildings with very high energy
performance, have become the standard in Europe in terms of building performance from
2021 onwards and benefiting from the recent Internet of Things (IoT) technological advances, they are equipped with on-site smart monitoring systems and services, advanced
communication technologies. The growing amount of data produced by such IoT applications has empowered decision-making via data analytics techniques.
This dissertation focuses on the development and comparison of predictive algorithms for
a state-of-the-art nZEB smart building and metered data produced by its sensors and advanced monitoring systems. In particular, this involves energy load consumption, as well
as weather data which are used to develop, train, and evaluate Machine Learning (ML)
models. Various ML algorithms and methodologies, as well as combinations thereof, are
explored and put to the test, each with its unique characteristics, in order to produce a
robust approach for One-Step-Ahead Energy Load Forecasting (OSA-ELF). The effect
of exogenous parameters is assessed, and comparisons are made between the different
methods tested on both variating portions of the training dataset but also on new unseen
data with similar properties and characteristics.
Collections
Show Collections
