Applications of artificial intelligence/machine learning to high-performance composites

Composites Part B Engineering, Journal Year: 2024, Volume and Issue: 285, P. 111740 - 111740

Published: July 23, 2024

Language: Английский

---

Machine Learning-Based predictions of crack growth rates in an aeronautical aluminum alloy

Theoretical and Applied Fracture Mechanics, Journal Year: 2024, Volume and Issue: 130, P. 104278 - 104278

Published: Jan. 9, 2024

Language: Английский

---

Mean stress correction and fatigue failure criteria for hyperelastic adhesive joints

The Journal of Adhesion, Journal Year: 2023, Volume and Issue: 100(4), P. 219 - 242

Published: May 25, 2023

The work investigates failure criteria and mean stress correction approaches for the fatigue lifetime prediction of two hyperelastic adhesives (a polyurethane, PU, a silicon-modified polymer, SMP). Fatigue experiments are carried under constant amplitude cyclic loading at RT 40°C/60% r.h with butt- thick-adherend-shear-test-joints three ratios R = −1, 0.1 0.5. Three evaluated: Goodman (static strength based), Schütz (mean sensitivity based) Kujawski & Ellyin (parameter optimisation based). considered are: Drucker-Prager (linear-relation hydrostatic stress), Beltrami (quadratic relationship multivariable nominal shear tensile stresses criterion (data-based). comparison is based on accuracy (R-squared master SN curves) complexity parameter determination. highest values R-squared were obtained by correction, followed then Goodman. However, determination follows an opposite trend being lightest approach. Failure yielded comparable results having advantage not dealing FEA, but limited to joints nearly uniform distribution. Finally, compared Drucker-Prager, had more robust

Language: Английский

---

Investigating permafrost carbon dynamics in Alaska with artificial intelligence

Environmental Research Letters, Journal Year: 2023, Volume and Issue: 18(12), P. 125001 - 125001

Published: Oct. 23, 2023

Abstract Positive feedbacks between permafrost degradation and the release of soil carbon into atmosphere impact land–atmosphere interactions, disrupt global cycle, accelerate climate change. The widespread distribution thawing is causing a cascade geophysical biochemical disturbances with impacts. Currently, few earth system models account for feedback (PCF) mechanisms. This research study integrates artificial intelligence (AI) tools information derived from field-scale surveys across tundra boreal landscapes in Alaska. We identify interpret cycling links sensitivities GeoCryoAI, hybridized multimodal deep learning (DL) architecture stacked convolutionally layered, memory-encoded recurrent neural networks (NN). framework in-situ measurements flux tower observations teacher forcing model training. Preliminary experiments to quantify, validate, forecast efflux Alaska demonstrate fidelity this data-driven architecture. More specifically, GeoCryoAI logs ecological memory effectively learns covariate dynamics while demonstrating an aptitude simulate PCF dynamics—active layer thickness (ALT), dioxide (CO 2 ), methane (CH 4 )—with high precision minimal loss (i.e. ALT RMSE : 1.327 cm [1969–2022]; CO 0.697 µ molCO m −2 s −1 [2003–2021]; CH 0.715 nmolCH [2011–2022]). variability sensitive harbinger change, unique signal characterizing PCF, our first characterization these space time.

Language: Английский

---

Workflow for fatigue life prediction of additive manufactured complex designs from powder bed fusion of Ti–6Al–4V

International Journal of Fatigue, Journal Year: 2023, Volume and Issue: 177, P. 107941 - 107941

Published: Sept. 9, 2023

Language: Английский

---

Fatigue life prediction of composite bolted joints based on finite element model and machine learning

Fatigue & Fracture of Engineering Materials & Structures, Journal Year: 2024, Volume and Issue: 47(6), P. 2029 - 2043

Published: March 26, 2024

Abstract This study proposes a fatigue life prediction method for composite bolted joints, which combines algorithm optimization‐based hybrid neural networks with finite element modeling. First, based on the Hashin failure criterion of physical mechanism, model joints is established, and simulation calculations have been conducted using various initial conditions. Then, by integrating experiment data, we established database that serves machine learning training prediction. Finally, data undergo comprehensive process deep feature extraction through utilization convolutional network (CNN). The resulting features are utilized as inputs backpropagation (BPNN) to predict life. results indicate this synergistic combination CNN BPNN in substantial improvement accuracy has remarkable superiority predicting joints.

Language: Английский

---

Dealing with multiaxial non-proportional fatigue with varying mean stress: invariant and critical plane approaches for component-like structural adhesive joints

International Journal of Adhesion and Adhesives, Journal Year: 2024, Volume and Issue: 134, P. 103782 - 103782

Published: July 24, 2024

A comparative investigation between an invariant-based approach using the signed Beltrami-stress and critical-plane Findley-stress is carried out for fatigue lifetime prediction of epoxy structural adhesive under multiaxial non-proportional loading with varying mean stress. Fatigue experiments uniaxial (proportional non-proportional) are done at stress ratios −1 0.1. Two joints evaluated: a hollow-cylinder butt-joint (HCBJ, sample homogeneous stresses distribution), flange-rod-joint (FRJ, component-like specimen complex state). For both geometries, increasing lead to strength reduction, whereas phase-shift effect less pronounced. systematic procedure parameter determination implemented approaches. predictions HCBJ-sample, analytically obtained. validation FRJ-sample, FEA-based calculation used. joints, critical plane results in more accurate than approach. In terms, experimentally complex, by requiring correction, but its numerically simpler lower time. The requires fewer provides loading. However, it demanding larger computing

Language: Английский

---

Design of Adhesive Bonded Joints

Processes, Journal Year: 2023, Volume and Issue: 11(12), P. 3369 - 3369

Published: Dec. 4, 2023

Adhesive bonded joints have become vital to modern engineering, offering advantages such as weight reduction, enhanced fatigue performance, and improved stress distribution [...]

Language: Английский

---

Compliance-Based Determination of Fatigue Design Curves for Elastomeric Adhesive Joints

Eng—Advances in Engineering, Journal Year: 2023, Volume and Issue: 4(4), P. 2615 - 2639

Published: Oct. 16, 2023

A compliance-based method for the determination of fatigue design curves elastomeric adhesive joints is developed and validated. Fatigue experiments are conducted on adhesives (a polyurethane a silane-modified polymer) under different stress ratios (R = 0.1/0.5/−1) conditions (23 °C/50% r.h. 40 °C/60% r.h.). The investigation focused butt thick adherent shear test joints. tests recorded with cameras to identify stages crack initiation propagation. For each test, stiffness compliance per cycle calculated until final failure. proposed identifies transition point that distinguishes regions stable unstable growth. then built based number cycles reach degrees initial (90%, 80%, 70% 60%). failure ratio, i.e., lifetime reaching given approach divided by total lifetime, introduced evaluate data in terms average values standard deviation. results indicate can yield high coefficient (accuracy) ratio (avoiding over-conservative design). Moreover, robust, as adhesives, ratios, geometries highly consistent.

Language: Английский

---

Transparent Hybrid Glass-Wood Bracing: Initial Results of an Experimental Campaign

Lecture notes in civil engineering, Journal Year: 2024, Volume and Issue: unknown, P. 521 - 536

Published: Oct. 31, 2024

Language: Английский

---