A Robust Fuzzy Model for Evaluating Defects in Building Elements

This study addresses the critical issue of building element defect management, focusing on prevalent concerns like cracks, dampness, and surface degradation. Recognizing the limitations introduced by human subjectivity in defect classification, the research proposes a novel, data-driven approach to automate the process. The methodology leverages extensive field data collection, encompassing 500 painted walls from 24 geographically dispersed buildings, to develop a robust fuzzy logic-based building element condition assessment model. The model categorizes element conditions (C1-C5) and classifies damage severity into five groups: no damage, slight damage, moderate damage, extensive damage, and complete damage, with nuanced precision. The efficacy of the fuzzy C-Means clustering is rigorously validated through the application of silhouette index and Davis-Bouldin index, ensuring optimal cluster formation and enhanced model accuracy. A real-world case study involving an office building exemplifies the model's practical application, showcasing its effectiveness in minimizing human error during defect identification and classification. This research contributes a sophisticated defect management framework informed by extensive field data and validated fuzzy logic, ultimately leading to demonstrably improved building quality and reduced operational costs within the construction industry.