Design framework and optimization of portable biomedical waste decomposition systems using ANN and MOPSO

Biomedical waste (BMW) incineration requires accurate prediction of energy demand and efficiency due to its heterogeneous composition. In this study, material and energy balance calculations were combined with design of experiments (DOE), analysis of variance (ANOVA), artificial neural network (ANN) modeling, and multi-objective particle swarm optimization (MOPSO). The novelty of the work presents the integration of metaheuristic optimization (MOPSO) into the biomedical waste incineration process. Results showed cellulose content as the most significant determinant of auxiliary energy requirement, with higher cellulose reducing LPG demand, while tissue and moisture exerted secondary but measurable effects. Efficiency ranged between 95.5 and 95.6%, with efficiency decreasing at higher moisture levels. The ANN model achieved near-perfect prediction accuracy (R² > 0.9999), enabling robust surrogate-based optimization. MOPSO analysis identified Pareto-optimal operating conditions where auxiliary energy demand reduced from 99.7 MJ/h to 97.2 MJ/h while efficiency improved from 95.52% to 95.60%. Under optimal waste composition identified by the ANN-MOPSO hybrid, auxiliary LPG consumption reduced from 33.8 to 27.4 kg/h, indicating strong potential for energy savings within the studied domain.