Machine learning-driven analysis of dielectric response in polymethyl methacrylate nanocomposites reinforced with multi-walled carbon nanotubes

This work investigates the complex dielectric spectroscopy of polymethyl methacrylate (PMMA) doped with non-functionalized, OH functionalized, and COOH functionalized multi-walled carbon nanotubes (MWCNTs) in a frequency range of 10 4 Hz to 2 MHz. Utilizing a 3D mixing technique, various MWCNT concentrations were reinforced in PMMA to create polymer nanocomposites, followed by injection compression. Extra tree regression analysis was then implemented to forecast properties such as dielectric constant, conductivity, loss tangent, and electric modulus at intermediate frequencies. To ensure robust model performance, training used subsets ranging from 50 to 70%, with the remaining 50 to 30% set aside for testing, respectively. Performance metrics such as adjusted R 2 score, root mean square error, and mean absolute error were employed to evaluate the predictive accuracy of the models. Experimental data obtained from tests highlighted that the application of extra tree regression analysis resulted in a noteworthy 50% reduction in both analysis time and associated resource expenditures. Novel insights into the chemical interactions and structural changes in the synthesized PMMA nanocomposites were gained through Fourier-transform infrared spectroscopy and X-ray diffraction techniques. This study not only demonstrates the efficiency of advanced regression techniques in predicting dielectric properties but also introduces a cost-effective approach for developing high-performance polymer nanocomposites. These findings have significant potential applications in electronics, sensors, and advanced materials, underscoring the novelty and practical relevance of the research.