Progressive collapse design of reinforced concrete frames using structural optimization and machine learning

Abstract

This paper uses an innovative algorithm combining machine learning as a decision-maker (DM) and Particle Swarm Optimization (PSO), called DMPSO, as a structural optimization technique, to design reinforced concrete frames for progressive collapse employing the alternate path method. In the alternate path method, multiple scenarios of removing critical elements should be considered, which makes the design process extremely repetitive and costly. Therefore, the development of an optimization technique is beneficial for producing efficient and cost-effective design solutions. The effectiveness of the proposed optimization algorithm is illustrated in optimization of a reinforced concrete structure that is subjected to lateral seismic forces, while the design concurrently satisfies both the American Concrete Institute provisions and the Unified Facilitates Criteria progressive collapse requirements. The results confirm the ability of the proposed DMPSO algorithm to efficiently find optimal design solutions in reinforced concrete structures that are subjected to progressive collapse.