[Article] Adaptive Neuro-Fuzzy Inference System-based Prediction of Heating and Cooling Loads in Residential Buildings

One of the greatest challenges in sustainable building design is making informed decisions during the earliest design stages, when architects have limited information but the greatest opportunity to influence future energy performance.

Our latest publication introduces an Adaptive Neuro-Fuzzy Inference System (ANFIS) that accurately predicts residential building heating and cooling loads using only a small set of passive design parameters.

Semahi, S., Benbouras, M. A., Zemmouri, N., & Attia, S. (2025). Adaptive Neuro-Fuzzy Inference System-based Prediction of Heating and Cooling Loads in Residential Buildings. Urbanism. Arhitectura. Constructii, 16(2), 225-251.

🔍 Why is this important?

Building energy simulations are powerful but often too time-consuming and data-intensive for conceptual design. Our work demonstrates that Artificial Intelligence can provide fast and reliable predictions without requiring detailed simulation models.

Using a calibrated EnergyPlus model of a representative Algerian multifamily residential building, we generated 1,200 simulation cases and trained two ANFIS models to predict both heating and cooling loads.

📊 Key findings

• The proposed ANFIS model achieved excellent predictive performance with R² = 0.90 for cooling loads and R² = 0.88 for heating loads, demonstrating strong agreement with dynamic simulation results.

• Sensitivity analysis revealed that only six passive design variables explain most of the variation in energy demand, including:

• Window-to-wall ratio
• Window thermal performance
• External wall insulation thickness
• Solar absorptance of external walls
• Building orientation
• Bedroom window area

• Transparent envelope characteristics, particularly glazing properties and window size, emerged as the dominant drivers of cooling demand, while insulation remained the most influential parameter for heating performance.

🌍 Why does this matter?

This work represents one of the first applications of ANFIS for residential building energy prediction in Algeria and contributes to the growing integration of Artificial Intelligence into building performance assessment across the MENA region. Rather than replacing detailed simulation, AI can support architects during the earliest design stages by rapidly exploring design alternatives before investing time in full dynamic simulations.

As the building sector moves toward net-zero emissions, combining building physics, sensitivity analysis, and machine learning will become increasingly important for accelerating evidence-based design decisions.

Congratulations to Samir Semahi, Mohammed Amin Benbouras, and Noureddine Zemmouri for this excellent collaboration.

📘 Full article: https://orbi.uliege.be/handle/2268/336120

📚 Learn more about our research https://www.sbd.uliege.be/

💻 Subscribe to my newsletter https://lnkd.in/diTVT5eq

🅱️ Bilibili b23.tv/bzjL3bn or 🎬 YouTube https://lnkd.in/erHrfkNf

🌐 Explore previous posts and resourceshttps://www.shadyattia.org

🔗 Follow all my professional links, including WeChat 🟩💬 微信https://lnkd.in/eN3xZhhZ

#ArtificialIntelligence #MachineLearning #ANFIS #BuildingPerformance #EnergySimulation #EnergyEfficiency #BuildingEnergy #BuildingDesign #SustainableBuildings #BuildingPhysics #EarlyDesign #ResidentialBuildings #EnergyPlus #BuildingScience #Architecture #NetZeroBuildings #ClimateAction #MENA #Algeria #UniversityOfLiege #SBDLab