Return
Influence of alkali treatment on interfacial adhesion and free vibration behavior of unidirectional carbon/ramie/epoxy hybrid laminates: an experimental, statistical and artificial neural network study
S
S
V
B
R
DOI:10.1080/01694243.2026.2696498.png)
Abstract
En 中文
Hybrid natural–synthetic composites are increasingly explored for lightweight vibration-sensitive structures; however, existing carbon/ramie hybrid composite studies mainly focus on static mechanical performance, while free vibration analysis, artificial neural network (ANN) prediction, and statistical validation are rarely integrated. In this study, a comprehensive experimental–statistical–computational methodology was developed to study NaOH-treated carbon/ramie/epoxy (CRE) hybrid laminates using vibration characterization, ANN-based prediction, and statistical analysis of the treated effect. Unidirectional CRE hybrid laminates were prepared with NaOH concentrations of 0, 1, 3 and 5 wt.%, and tested under clamped-free (CF) and clamped-clamped (CC) boundary conditions with a calibrated impulse hammer–accelerometer system (uncertainty ±1.5 Hz). The optimal vibration response was obtained for 3 wt.% NaOH (CR2CE03); the CF Mode 1 frequency increased from 35.8 Hz to 42.5 Hz, and the CC Mode 3 frequency increased from 935.6 Hz to 1413.7 Hz, with a decrease in damping caused by an increase in interfacial bonding. The relationship between flexural modulus and natural frequency was determined by Euler – Bernoulli beam theory and Pearson correlation analysis (r = 0.934–0.971). These results were confirmed by ANOVA which showed a significant treatment effect (p < 0.05) and very large effect size (η2 = 0.593–0.989). A Levenberg–Marquardt ANN model was developed that was highly accurate in predicting the frequency (R = 0.99394) and the damping (R = 0.97046), which was validated using Root Mean Square Error and Mean Absolute Error.
Keywords:
Carbon/ramie/epoxy hybrid composites
free vibration analysis
natural frequency
damping factor
alkali surface treatment
artificial neural network and one-way ANOVA analysis
Journal
J
IF:
3.7
Papers:
340
Citations:
6.8K
