Biomass Co-Pyrolysis Optimization

Clean energy from biomass-coal co-pyrolysis sounds straightforward—mix and burn. But the non-linear interactions between mixing ratios and gas/char/liquid yields make brute-force optimization impossible. We built a three-model ensemble: LightGBM for yield prediction, Gaussian Process Regression for uncertainty quantification, and Particle Swarm Optimization to search the non-convex solution space.

The entropy-weighted fuzzy evaluation framework captured interactions that single-objective models miss entirely. PSO converged on 28.44% as the optimal biomass-to-coal ratio—a specific, actionable number that maximizes clean energy output. Judges at Shuwei Cup praised the methodology as transferable across energy domains, awarding First Prize plus the Grand Innovation Award.