Strength estimation of textured polymer layer-reinforced materials in practical marine engineering based on physical experiments and artificial intelligence modellingShow others and affiliations
2025 (English)In: Frontiers in Marine Science, E-ISSN 2296-7745, Vol. 12, article id 1653741
Article in journal (Refereed) Published
Abstract [en]
Marine coral sand-clay mixtures (MCCM) are widely used as fill materials in ocean engineering, where their strength is influenced by marine clay content. This study investigates the mechanical behavior of textured polymer layer-reinforced MCCM using 3D-printed technology with varying asperity heights, spacings, and reinforcement layers. Triaxial tests reveal that increased reinforcement, higher asperities, and smaller spacings enhance strength and internal friction angle with minimal effect on cohesion. Particle breakage increases with reinforcement, and fractal analysis shows a linear relationship between fractal dimension and breakage rate. SEM images reveal the complex interfacial interaction mechanisms between the MCCM and the polymer layer. A comprehensive dataset from these tests supports the development of predictive models, including BPNN, GA-BPNN, PSO-BPNN, and LDA-BPNN, with the LDA-BPNN showing the highest accuracy and generalization. Compared with existing approaches, the proposed model framework achieves significant improvements in predictive performance and robustness. Sensitivity analysis identifies asperity spacing and asperity height as key factors. An empirical formula derived from the LDA-BPNN enables practical strength prediction, offering valuable guidance for marine construction design.
Place, publisher, year, edition, pages
Frontiers Media S.A., 2025. Vol. 12, article id 1653741
Keywords [en]
3D printing technology, machine learning, marine coral sand-clay mixture, textured polymer layer reinforcement, triaxial shear tests
National Category
Condensed Matter Physics
Identifiers
URN: urn:nbn:se:umu:diva-243757DOI: 10.3389/fmars.2025.1653741ISI: 001555352000001Scopus ID: 2-s2.0-105014004765OAI: oai:DiVA.org:umu-243757DiVA, id: diva2:1994954
2025-09-042025-09-042025-09-04Bibliographically approved