Analytical Study on Strength and Durability of Self-curing Concrete Using Deep Learning

Self-curing concrete is a type of concrete that can automatically maintain adequate moisture and hydration without external curing methods, enhancing construction efficiency and durability. The workflow of this paper involves data collection, pre-processing, feature extraction, feature selection using a hybrid optimization model, and the development of a hybrid deep learning model combining GRU and CNN. Data collection involves gathering self-curing concrete parameters from literature sources, followed by pre-processing techniques like data-by-data handling (missing data removal) and min-mx normalization to clean the data and ensure consistent feature scales. Statistical features are extracted to provide quantitative information about the dataset, and feature selection is performed using a hybrid optimization model that combines the Ant Lion Optimizer (ALO) Algorithm and Crow Search Algorithm (CSA) to identify the most relevant features for strength and durability analysis. The hybrid deep learning model combines GRU and CNN components to capture temporal dependencies, sequential information, and spatial features. The final prediction is obtained through a voting approach. This approach improves the understanding of self-curing concrete’s strength and durability characteristics, providing accurate predictions and insights for optimizing its design and performance in construction applications.