Optimizing rural building design with an intelligent framework integrating BES ANN and MCDM

Area of Science:

• Building Science
• Sustainable Architecture
• Computational Design

Background:

• Rural buildings face challenges in balancing energy consumption, thermal comfort, and construction costs.
• Existing optimization methods often struggle with high-dimensional design spaces, leading to slow convergence and computational complexity.

Purpose of the Study:

• To propose and validate a novel performance optimization framework for rural buildings.
• To integrate Building Energy Simulation (BES), Artificial Neural Networks (ANN), and Multi-Criteria Decision-Making (MCDM) for multi-objective optimization.
• To improve prediction accuracy and efficiency in the design of energy-efficient rural structures.

Main Methods:

• The framework integrates DesignBuilder and JePlus for batch simulations, using the Morris method for sensitivity analysis.
• An Artificial Neural Network (ANN) model was developed in MATLAB for performance prediction.
• The TOPSIS approach was employed within the Multi-Criteria Decision-Making (MCDM) component to select optimal design solutions.

Main Results:

• The optimized model demonstrated a significant reduction in total energy consumption by 61.64% and discomfort hours by 32.04%.
• The optimization was achieved with an additional construction cost of ¥73,519.6, indicating a well-balanced performance improvement.
• The proposed framework showed enhanced prediction accuracy and optimization efficiency compared to conventional methods.

Conclusions:

• The developed BES-ANN-MCDM framework offers a novel and effective approach for optimizing rural building design.
• The study provides a replicable pathway and theoretical foundation for performance-driven, energy-efficient rural building design.
• The findings highlight the potential for substantial improvements in energy efficiency and thermal comfort in rural housing.