A study by the University of La Laguna (ULL) has developed machine learning models to predict hotel energy demand 24 hours in advance. This research, published in the Logic Journal of the IGPL, aims to optimize resource management, reduce energy consumption, and decrease greenhouse gas emissions in the hotel sector, one of the main economic drivers of the Canary Islands.
The work, titled Towards smart hotels: energy forecasting with machine learning models, is the result of collaboration by the Control Engineering and Intelligent Systems group at ULL. The proposed methods, such as Long Short Term Memory (LSTM) and Gated Recurrent Unit (GRU), are designed to process sequential data and have shown improved prediction accuracy when incorporating real-time data.
The method's evaluation was conducted in a luxury hotel in Tenerife. The results indicate that these precise prediction models can significantly reduce operational costs and maximize the use of renewable energy. Proper planning of electrical loads allows for the utilization of surplus energy, reduction of demand peaks, and optimization of available energy resources.
Researchers point out that models predicting 24 hours of demand simultaneously achieve lower error rates. Specifically, a 7.1% reduction in load shifting and a 23.5% reduction in load accommodation were observed, optimizing facility operations. These tools are key to advancing towards more efficient and sustainable hotels.
The ULL team plans to adapt the model to other establishments and areas in the Canary Islands with different climatic conditions. These methods are part of the Interreg Atlantic Area SATCOMM project, in which the university collaborates with ten other European entities.
