Machine Learning Prediction of the Yield and BET Area of Activated Carbon Quantitatively Relating to Biomass Compositions and Operating Conditions

Althoughactivated carbon's yield (quantity index) and BETarea (quality index) are crucial to its application, the two indexesmust be accurately predicted. Herein, biomass compositions (ultimateanalysis, proximate analysis, and chemical analysis), operating conditions(mass ratio, carbonization time, carbonization temperature, activationtime, and activation temperature) under physical activation (CO2 and steam), and chemical activation (H3PO4, KOH, and ZnCl2) conditions as input parameterswere used to predict the two indexes of activated carbon simultaneouslythrough the random forest (RF) method for the first time. In total,the samples (>1500 data) identified from experiments in the literaturewere used to train, validate, and test the RF models. The resultsshow that the model built on ultimate analysis is more suitable forpredicting the BET area and yield of activated carbon prepared byboth physical and chemical activation. Therein, the R (2) values of activated carbon's yield and BET areaunder the H3PO4 activation condition were thehighest, which were 0.98 and 0.97, respectively. In addition, theinfluence of various factors and interactions on the target variableswas analyzed. The results show that the hydrogen content has a largeimpact on the yield under physical activation conditions, and themass ratio has the most contribution to the BET area under chemicalactivation conditions. This study affords achievable hints to thequantitative prediction of porous materials affected by multiple compositionsof raw materials and different operating conditions.