Machine Learning Based Prediction of Energy Generation in Plant–Microbial Fuel Cells Using Environmental and Soil Data

Abstract

Plant–Microbial Fuel Cells (P-MFCs) are emerging as sustainable, bio electrochemical systems that harness energy from the interaction between plant root exudates and microbial communities in soil. Predicting energy generation from P-MFCs is challenging due to the influence of multiple environmental factors (temperature, moisture, pH, soil nutrients, etc.). This study proposes a machine learning (ML) framework that predicts electrical output in P-MFCs using environmental and soil sensor data. We evaluated multiple ML models including Random Forest (RF), Support Vector Regression (SVR), and Gradient Boosting Machines (GBM) using data collected from field trials and laboratory experiments. Results show that ensemble methods (RF and GBM) achieve strong predictive accuracy (R² > 0.92) across varied environmental scenarios, outperforming SVR. Feature importance analysis reveals soil moisture and pH as key predictors of power output. The proposed model enables real time prediction and optimization of P-MFC energy generation, offering a path toward smarter bioenergy harvesting from agricultural ecosystems.