
Afsoun Abbasi
AI algorithms applied to the Energy Center plant data to determine the thermal consumption of the structure. Validation of the algorithms on real consumption data.
Rel. Davide Papurello. Politecnico di Torino, Corso di laurea magistrale in Data Science And Engineering, 2025
Abstract: |
Analisi dati da multisensore ambientale Abstract “At the Energy Center plant in Turin, measuring, understanding, and predicting thermal energy consumption plays a crucial role in improving operational efficiency and supporting sustainable energy use. This thesis investigates data-driven methods for forecasting thermal energy demand, using real sensor data collected from the plant over a five-month period between October 2024 and March 2025. The dataset includes high-resolution temperature readings from multiple points in the district heating network, along with cumulative energy consumption measurements. To make the data suitable for predictive modeling, several preprocessing steps were applied: aligning time series with different sampling rates, interpolating missing values, filtering out anomalies, and performing feature engineering. Among the features created, a key one “Virtual Energy” was developed by combining pump activity and temperature differentials, providing a more insightful representation of heat transfer within the system. A variety of machine learning models were tested, with particular focus on XGBoost and Long Short-Term Memory (LSTM) neural networks. After tuning and training, the best LSTM model—trained on smoothed time-series data—achieved excellent performance, with an R² score of 0.97 and a mean absolute error of approximately 2,100 for 15-minute-ahead predictions. Even when forecasting an hour ahead, the model maintained a high degree of accuracy and robustness. The study demonstrates that combining sensor data with advanced machine learning algorithms is a promising approach to short-term thermal energy forecasting. These models not only capture complex temporal patterns but also enable more proactive energy management. The results support the use of predictive modeling in district heating systems for real-time monitoring, performance optimization, and the development of intelligent control strategies.” |
---|---|
Relatori: | Davide Papurello |
Anno accademico: | 2024/25 |
Tipo di pubblicazione: | Elettronica |
Numero di pagine: | 76 |
Informazioni aggiuntive: | Tesi secretata. Fulltext non presente |
Soggetti: | |
Corso di laurea: | Corso di laurea magistrale in Data Science And Engineering |
Classe di laurea: | Nuovo ordinamento > Laurea magistrale > LM-32 - INGEGNERIA INFORMATICA |
Aziende collaboratrici: | NON SPECIFICATO |
URI: | http://webthesis.biblio.polito.it/id/eprint/36343 |
![]() |
Modifica (riservato agli operatori) |