Abstract

The transition toward sustainable waste-to-energy systems necessitates the efficient degradation of recalcitrant
lignocellulosic biomass. This study investigates the biomethane potential (BMP) of sugarcane bagasse (SB),
maize cob (MC), and coconut husk (CH) through a novel bio-inspired approach utilizing termite-derived
degradation The primary objective was to evaluate the synergistic effects of termite-mediated enzymatic
breakdown on methane yield and to develop a robust computational framework for process prediction. To
achieve this, an Artificial Neural Network (ANN) using a multi-layer perceptron (MLP) architecture was
designed and optimized.
Experimental data from termite-derived degraded agro solid waste trials served as the basis for the model,
which utilized a Levenberg-Marquardt back-propagation algorithm. To ensure statistical rigor and eliminate
stochastic bias, a sensitivity analysis was performed through a multi-run optimization loop across 2 to 20
hidden neurons. Results indicated that termite digestion activities significantly enhanced the degradation of
high-lignin substrates, particularly coconut husk, which typically exhibits high resistance in conventional
systems. The optimized ANN model, featuring 2 - 6 hidden neurons, demonstrated superior predictive
performance with a Coefficient of Determination (R2) exceeding 0.98 and 0.33 minimal Root Mean Square
Error (RMSE).
The findings confirm that termite-mediated degradation effectively reduces the structural recalcitrance of
agricultural solid wastes, while the ANN framework provides a highly accurate tool for simulating non-linear
anaerobic digestion kinetics. This research offers a scalable strategy for optimizing lignocellulosic
bioconversion, bridging the gap between bio-inspired catalysis and industrial-scale energy production.