Assessment of Concrete Structures Using a Combination of Field Tests and Artificial Intelligence Algorithms (Concrete DNA System)

Concrete structures in Libyan coastal environments pose a significant challenge due to harsh climatic conditions, where high humidity and salt-laden winds reduce the service life of these structures by up to 40% compared to their counterparts in inland areas. Conventional assessment methods face substantial difficulties in the accurate detection and analysis of hidden cracks, increasing the risk of collapse and necessitating the development of innovative solutions. This research presents the intelligent "Concrete DNA" system, which integrates field testing and artificial intelligence techniques to assess the condition of concrete structures. The system relies on the integration of two main algorithms: a K-Nearest Neighbors (KNN) algorithm for processing and classifying crack images, and an Artificial Neural Network (ANN) to fuse these results with data from field Schmidt Hammer tests. The system was trained on a dataset of 158,000 images from the IEEE Data Port and Kaggle repositories, where advanced image processing techniques, including Gaussian filtering and edge detection analysis, were applied. The results showed a diagnostic accuracy of 98% in identifying defects, with the capability to generate repair recommendations based on ACI/ASTM standards within just 3 seconds, compared to traditional methods. The system was successfully implemented on the Civil Engineering Department building at the University of Sabratha, demonstrating high efficiency in monitoring cracks and estimating compressive strength with an accuracy of ±0.2 N/mm². The system also achieved notable savings, reducing maintenance costs by 40% and increasing engineer productivity twenty-fold.