Energy utilization of agricultural waste: Machine learning prediction and pyrolysis transformation

Abstract

The rapid screening of agricultural waste materials for capacitor preparation holds significant importance in comprehending the relationship between material properties and enhancing experimental efficiency. In this study, we developed two machine learning models to predict electrode material characteristics using 2997 data points extracted from 235 articles. The identification and influence of key features on prediction indices provide a theoretical foundation for subsequent practical preparation. Through regression analysis and index evaluation, corn straw emerged as the optimal material for capacitor preparation, leading us to propose a one-step activation and two-step modification approach to convert corn straw into porous biochar. By modifying biochar with Co(NO₃)₂·6H₂O, the maximum electrode capacitance of porous carbon reached 732.6 F/g. Furthermore, the electrode exhibited exceptional cycle stability with a remaining capacitance of 96 % after 5000 cycles. The prepared symmetric capacitor demonstrated pseudocapacitance behavior with a capacitance of 183.15 F/g at a current density of 1.0 A/g, power density of 22 kW/kg, and energy density of 9.03 Wh/kg. Considering the increasing annual output of corn straw and its superior industrial application prospects compared to acid-, base-, or precious metal-based alternatives due to their cost-effectiveness and environmental friendliness, these findings highlight the potential practical value in utilizing modified corn straw biochar as an efficient energy storage electrode material.

Keywords: Agricultural waste; Energy storage; Machine learning; Pyrolytic activation.