A LIME-LSTSNM approach based green building sustainability prediction using BIM design

This research presents a climate change-based parameter optimisation approach for sustainable green building design. The process begins with a Building Information Modeling (BIM)-based design, followed by a Design-Builder simulation. Climatic data is collected and pre-processed, and building parameters are optimized using SA2O, considering this data. BIM-based building parameters and the optimized data are then extracted. The simulation output, along with sensor and historical data, are fused using the Multiresolution Kalman Filter (MKF) technique. Incomplete data is handled with Penalized K-Log Euclidean Neighbor (PKLEN), followed by season-based grouping using KMA. Non-linear dynamics are analyzed, and features are extracted from both the grouped and non-linear data. The sustainability factor is predicted using Local Interpretable Model-agnostic Explanations (LIME), with Long Short-Term Skip Norm Memory (LSTSNM), and feedback is provided to optimise the building parameters for sustainable green building design. Experimental results show that this model achieved an accuracy of 98.24 %, demonstrating the effectiveness of the proposed approach in enhancing sustainability in building design while considering climate change.