Optimization of reinforced concrete structures using AI

Abstract

Optimization of reinforced concrete structures has traditionally relied on mathematical methods to find rational combinations of geometric parameters, material characteristics, and reinforcement while meeting strength, rigidity, and structural suitability requirements. The development of artificial intelligence and machine learning significantly expands the capabilities of this approach. Modern algorithms allow not only to search for optimal design solutions but also to approximate the results of labor-intensive calculations, generate surrogate models, predict structural characteristics, and combine multiple optimality criteria into a single procedure. This research article provides an analytical review of methods for applying artificial intelligence to the optimization of reinforced concrete structures. It examines genetic algorithms, neural networks, machine learning methods, metamodels, hybrid algorithms, and deep reinforcement learning. Particular attention is paid to multi-criteria optimization, computational cost reduction, environmental performance, and the need for regulatory verification of results. The scientific novelty of the article lies in the proposed conceptual model of AI-oriented optimization of reinforced concrete structures, which combines a digital model of the structure, a computational database, a ML surrogate, an intelligent optimizer, and independent finite element and regulatory verification into a single iterative process.