Treffer: Unveiling predictive insights for enhanced performance of PVDF-based nanogenerators via machine learning modeling.

[Display omitted] • ML models to optimize triboelectric nanogenerator (TENG) fabrication. • Evaluation of 3 ML models (DTR, RF, GBR) for TENG voltage optimization. • GBR outperforms with high R2 values: Train 0.9812; Test 0.9370. Triboelectric nanogenerator (TENG) based on polyvinylidene fluoride (PVDF) have demonstrated substantial potential in the domain of mobile energy generation, particularly for self-powering ultra-low-voltage microdevices. The current research presents a comparative analysis and guided-approach study for fabricating PVDF-based TENG with voltage predictive capability using predictive machine learning (ML) models. In this study, three distinct ML algorithms, specifically Decision Tree Regression (DTR), Random Forest (RF), and Gradient Boosting Regression (GBR), were employed to optimize the fabrication process and improve upon the existent voltage output performance of PVDF nanogenerators. Experimental data was collected for various TENG fabrication parameters, including material properties, configurations, and processing methods. The GBR model outperformed the other two models, exhibiting the highest R2 value of 0.9812 and 0.9370 for train and test, respectively. To facilitate practical implementation, a user-friendly Python-based graphical user interface was developed, enabling researchers to utilize the predictive models for real-time voltage output analysis. This study strategically utilized PVDF as the principal material and electrospinning as the preferred fabrication method to minimize potential variables and errors, thereby allowing for a targeted exploration of TENG's specific aspects. The preliminary findings are expected to lay the groundwork for enhanced comprehension and fine-tuning of TENG's efficacy. [ABSTRACT FROM AUTHOR]