Indoor microclimate assessment for cultural heritage buildings: Statistical analysis and machine learning-based prediction

Preserving historic buildings of significant cultural value requires balancing the often-competing demands of energy efficiency, heritage conservation, and occupant comfort. This work investigates the indoor microclimate of the 12th-century Church of Santa Maria Annunziata in L’Aquila, Italy, which houses important artworks, including paintings and wooden objects. The study focuses on the main indoor microclimatic parameters: temperature, relative humidity, and illuminance. Currently lacking a dedicated HVAC system, the building’s indoor environmental stability was assessed using a multi-disciplinary approach that integrated experimental measurements, statistical analysis, and predictive algorithms across varying time horizons. The results indicate that current thermal conditions frequently fluctuate outside the established threshold for proper art preservation; specifically, 55.4% of temperature readings fell outside the 19-24 ∘C range, 87.6% of relative humidity values fell outside the 55–60% range, and 19.1% of illuminance values fell outside the 0–50 lx range. The Performance Index (PI) for temperature and relative humidity consistently revealed critical values, reaching a mere 6.73% even in the best-case scenario. A predictive analysis was conducted on temperature and relative humidity using Artificial Neural Networks, Regression Trees, and Random Forests across multiple forecasting horizons (30 minutes to 6 hours). The results demonstrate high predictive accuracy for both short-term and long-term horizons, highlighting the potential of data-driven models to support future decision-support systems and predictive strategies aimed at enhancing the conservation management of historic buildings.