Microwave Pyrolysis of Rapeseed Straw: Optimization and Prediction of Solid Biofuel Properties Using RSM and ANN-GA

Biomass energy utilization can address the growing demand for renewable energy while alleviating environmental pressures. Pyrolysis has emerged as a key thermochemical conversion technology for upgrading biomass into clean and efficient bioenergy products. In this study, response surface methodology (RSM) and artificial neural network-genetic algorithm (ANN-GA) were employed to optimize the process parameters for microwave vacuum pyrolysis of rapeseed straw. A central composite design (CCD) was applied to investigate the effects of three independent variables, namely microwave power (200–400 W), irradiation time (10–30 min), and absorber dosage (3–15 g). The optimization objective was to maximize an entropy-weighted composite score of solid yield, higher heating value (HHV), and energy-mass co-benefit index (EMCI). The Bayesian regularization (BR)-trained ANN model outperformed RSM in predictive accuracy, yielding higher R2 and lower RMSE. The optimal ANN-GA conditions were 325.16 W, 20.62 min, and 5.43 g, which were experimentally validated to achieve a solid yield of 77.24%, an HHV of 18.07 MJ/kg, and an EMCI of 21.33%. These results confirmed that the hybrid ANN-GA framework is an effective approach for optimizing biomass pyrolysis processes to produce high-quality solid biofuels.