Machine Learning Approaches for Mechanical Rebar Couplers: A Comprehensive Review of Alternatives to Lap Splices

Mechanical rebar couplers have become a traditional lap splicing in reinforced concrete designs, especially in those areas where reinforcement congestion and high seismic performance are strict. They have a mechanical performance that is dependent on a variety of parameters that are based on material properties, geometry, installation quality, and loading conditions, and their prediction and optimization are challenging under conventional analytical methods. This paper is a critical and detailed review of machine learning methods for the mechanical rebar couplers with an accent on their usage as alternatives to lap splices in modern-day reinforced concrete construction. Types of mechanical couplers, parameters of mechanical and structural performance governance, sources of experimental data and numerical data, and state-of-the-art ML algorithms that allow us to predict the behavior of mechanical couplers are systematically discussed in the review. The current developments, such as hybrid and physics-informed machine learning frameworks, are touched upon in the context of structural reliability, design optimization, and sustainability. The research gaps are recognized, and the main areas include the standardization of datasets, explained models of machine learning, and enhanced correlation between machine learning predictions and structural design codes. This review will also provide a consolidated base of knowledge and a roadmap of future research to researchers and practitioners in intelligent design and evaluation of mechanical rebar couplers.