Comparison of Transfer Learning Techniques for Building Energy Forecasting
Shansita Das Sharma, Austin Coursey, Marcos Quinones-Grueiro, and 1 more author
IFAC-PapersOnLine, 2024
12th IFAC Symposium on Fault Detection, Supervision and Safety for Technical Processes SAFEPROCESS 2024
The growing demand for building energy efficiency necessitates accurate predictions of normal versus abnormal operations to understand their impact on energy management. However, integrating predictive models into practical applications faces challenges, especially in buildings with limited measurements and data. This paper explores the viability of three widely adopted transfer learning techniques in improving energy consumption models, focusing on real-world data with internal building measurements. The findings suggest that transferring information between buildings is a promising method to provide positive improvements in energy prediction models.