Research on Computer-Aided Architectural Design Optimization Based on Building Information Modeling (BIM) Technology

This study explores the integration of Building Information Modeling (BIM) technology in computer-aided architectural design optimization, aligning with advancements in computational science and sustainable engineering. Architectural design optimization traditionally involves balancing cost, performance, and sustainability under complex constraints. Existing methods, such as gradient-based optimization and heuristic algorithms, often face limitations in scalability and adaptability, particularly when addressing conflicting objectives and dynamic constraints. To overcome these challenges, this research introduces the Adaptive Architectural Design Network (AADNet), a hybrid framework combining deep learning with optimization algorithms. AADNet incorporates domain-specific knowledge through its Constraint-Adaptive Representation Module (CARM), enabling dynamic adjustments to complex architectural constraints, and the Optimization Refinement Layer (ORL), which iteratively improves design solutions using gradient-based heuristics. Furthermore, the Context-Aware Optimization Strategy (CAOS) enhances AADNet by employing dynamic objective balancing, multi-scale feedback, and adaptive regularization, ensuring robustness and scalability. Experimental results demonstrate significant improvements, achieving up to 94.78% accuracy on the ShapeNet dataset and a 91.32% Intersection over Union (IoU), showcasing rapid convergence and effective handling of multi-objective trade-offs, with significant enhancements over existing methods in both computational efficiency and design optimization.