Simulation and optimization of the post plasma-catalytic system for toluene degradation by a hybrid ANN and NSGA-II method

In this study, a post-non-thermal plasma (NTP)-catalytic system was developed for the removal of toluene over a series of MnCoO_x/γ-Al₂O₃ catalysts. The addition of the MnCoO_x/γ-Al₂O₃ catalysts markedly promoted the toluene removal efficiency, CO_x yield, CO₂ yield and energy yield (EY) compared with the plasma alone system. The 5 wt% MnCoO_x/γ-Al₂O₃ catalyst exhibited the best reaction performance, which could be attributed to the reducibility and surface active oxygen species of the catalyst. With artificial neural network (ANN), the effects of experimental parameters on the reaction performance of toluene degradation were modeled and analyzed; for this analysis, four parameters were considered, namely, discharge power, initial concentration of toluene, flow rate, and relative humidity. The results indicated that the predicted results fitted well with the experimental results. The discharge power was the most significant factor for the toluene removal efficiency and CO_x yield, whereas the EY was the most influenced by the gas flow rate. A multi-objective optimization model was proposed to determine optimal experimental parameters, which was then solved using the non-dominating sorting genetic algorithm II (NSGA-II). The results revealed that the Pareto front obtained from the hybrid ANN and NSGA-II method provided a series of feasible and optimal process parameters for the post-NTP-catalytic system. This hybrid method also served as an effective tool to select process parameters according to application conditions and preferences.