Energy Consumption Prediction Using Recurrent Neural Network (LSTM) : Multivariate Time Series forecasting
Nachawati, Mohamad (2023)
2023
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2023121035798
https://urn.fi/URN:NBN:fi:amk-2023121035798
Tiivistelmä
Energy consumption prediction plays a crucial role in optimizing resource allocation and ensuring efficient energy management. This thesis delves into the realm of multivariate time series forecasting, with a specific focus on leveraging Recurrent Neural Network (RNN) architecture, particularly Long Short-Term Memory (LSTM) models.
The investigation places a strong emphasis on the importance of meticulous data preprocessing and adept feature engineering in the energy consumption prediction process. The utilization of the LSTM model is a key focal point, with an in-depth exploration of its development and refinement throughout the project. Drawing data from reputable sources such as Fingrid, the Finnish Meteorological Institute, and the World Bank, the study employs advanced visualization techniques to extract meaningful insights. The LSTM model demonstrates remarkable predictive capabilities, as evidenced by consistently low mean absolute error, mean squared error, and root mean squared error values.
In summary, this study underscores the effectiveness of LSTM models for energy consumption prediction, especially in capturing short-term dynamics. Despite challenges in predicting long-term trends, the findings showcase the potential of LSTM models in optimizing energy management. The research concludes by proposing avenues for future exploration, including the investigation of alternative model architectures and the real-time implementation of these models in diverse contexts.
The investigation places a strong emphasis on the importance of meticulous data preprocessing and adept feature engineering in the energy consumption prediction process. The utilization of the LSTM model is a key focal point, with an in-depth exploration of its development and refinement throughout the project. Drawing data from reputable sources such as Fingrid, the Finnish Meteorological Institute, and the World Bank, the study employs advanced visualization techniques to extract meaningful insights. The LSTM model demonstrates remarkable predictive capabilities, as evidenced by consistently low mean absolute error, mean squared error, and root mean squared error values.
In summary, this study underscores the effectiveness of LSTM models for energy consumption prediction, especially in capturing short-term dynamics. Despite challenges in predicting long-term trends, the findings showcase the potential of LSTM models in optimizing energy management. The research concludes by proposing avenues for future exploration, including the investigation of alternative model architectures and the real-time implementation of these models in diverse contexts.