Hybrid Intelligence approach to study post-processing impact on the mechanical performance of notched additively manufactured AlSi10Mg

This study introduces a Hybrid Intelligence approach to investigate the Process-Structure-Property-Performance (PSSP) relationship in additively manufactured (AM) materials, specifically focusing on V-notched laser powder bed fused (L-PBF) AlSi10Mg specimens. The Humen Intelligence (HI) component managed the design, manufacturing processes, post-processing, structural characterization, mechanical testing, and data collection. In parallel, Artificial Intelligence (AI), utilizing advanced machine learning (ML) algorithms, performed tasks related to prediction, sensitivity analysis, and parametric analysis. AI identified patterns and developed predictive models that provided deeper insights into how process parameters affect material properties and performance. This integration of HI and AI enabled a more thorough exploration of these relationships; data collected from our previous research were complemented with new experiments conducted to assess the effects of various heat treatments (HTs) and surface post-treatments (SPTs) on the fatigue behavior of the specimens. The techniques applied included stress relief (SR), T6 thermal treatments, sand blasting (SB), shot peening (SP), severe vibratory peening (SVP), laser shock peening (LSP), tumble finishing (TF), abrasive flow machining (AFM), chemical polishing (CP), electrochemical polishing (ECP), and chemical milling (CM), along with their combinations. A total of 54 different post-processing techniques were examined in this study. The experimental data, covering surface texture, microstructure, porosity, hardness, and residual stress, were used to develop an ML model that analyzed the fatigue behavior of the specimens. This approach represents a significant advancement toward integrated mechanistic and data-driven materials engineering, offering valuable insights for optimizing fatigue performance in practical applications.