Hyperparameter tuned machine learning predictions of specific capacitance of conducting polymers and their composites for high performance advanced supercapacitors

Abstract

This research investigates the use of machine learning (ML) to improve the performance of conducting polymer-based electrodes in supercapacitors, which leverage both electric double-layer capacitance (EDLC) and pseudocapacitive characteristics. Six ML models—Support Vector Machine (SVM), Random Forest (RF), Multilayer Perceptron (MLP), Decision Tree (DT), Extreme Gradient Boosting (XGBoost), and Artificial Neural Network (ANN)—are evaluated for their ability to predict the specific capacitance of electrodes using an experimental dataset comprising Polyaniline (PANI), Polypyrrole (Ppy), and Polythiophene (PTh). Performance metrics included Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared (R²). Among the models, MLP demonstrates superior predictive accuracy, achieving the lowest MAE of 0.1452 and MSE of 0.0373, along with the highest R² of 0.9622. In contrast, Decision Tree and SVM exhibited higher error values, with MAEs of 0.2107 and 0.2267 and R² values around 0.885. Although Random Forest and XGBoost achieved competitive R² values of 0.9399 and 0.9354, their errors are comparatively higher than MLP. These results highlight the effectiveness of advanced ML techniques in enhancing supercapacitor technology and indicate the potential of these models to predict and optimize conducting polymer-based electrode materials for improved performance.