Advancing Severe Plastic Deformation for Tubular Samples: Systematic Review and Machine Learning Optimization

Abstract

ABSTRACT Severe Plastic Deformation (SPD) is a widely recognized method for producing ultrafine‐grained (UFG) and nano‐structured materials with superior mechanical properties by imposing high strains without the need for alloying elements or secondary reinforcements. Although SPD techniques for bulk and sheet materials have been extensively developed, recent advancements have focused on creating effective SPD methods specifically designed for tubular samples. These techniques are critical for industries where lightweight, high‐strength tubular components are essential, such as aerospace, automotive, and biomedical sectors. Machine learning (ML) approaches have emerged as powerful tools for optimizing SPD parameters and predicting material behavior. ML involves constructing systems capable of analyzing and identifying patterns in data to make informed decisions, such as supervised and unsupervised learning, which analyze training data to reveal relationships and trends. When dealing with small datasets, cross‐validation (CV), particularly k‐fold CV, is a reliable technique to prevent overfitting and enhance model robustness. ML and statistical analysis were employed to predict and optimize the input parameters in terms of the plastic strain (PS) and the impact on the improvement of the mechanical properties. This review provides a comprehensive analysis of SPD processes tailored for tubular materials, including methods such as High‐Pressure Tube Twisting (HPTT), Accumulative Spin Bonding (ASB), and Parallel Tubular Channel Angular Pressing (PTCAP). Each method is examined in terms of deformation behavior, die design, and microstructural evolution. Additionally, the review addresses critical research gaps, explores potential industrial applications, and highlights how advancements in ML can enhance the understanding and application of SPD techniques, offering insights into scaling up these processes for large‐scale production.