基于过程模拟和随机森林模型的生物质制氢过程因素分析与预测

Biomass can replace fossil fuels, reduce greenhouse gas emissions, and is a promising renewable energy source. Co-production of multiple-products has been demonstrated efficient and economically viable process. The techno-economic feasibility of biomass conversion into hydrogen (H₂) and activated carbon (AC) route has also been analyzed. However, the selection of raw materials and process parameters become the main barrier for scale-up production. The different types of biomass species and the process conditions affect the yield and quality of the products. In this paper, a process model was developed to simulate the biomass conversion process. H₂was produced from biomass through pyrolysis and chemical looping gasification processes. AC was fabricated from biomass through carbonization and activation processes. Then, machine learning was used to build a high-quality prediction model and accordingly explore the importance factors in producing demanded products. The results indicated that process parameters had greater influence than the raw materials on H₂ concentration and yield. For example, hydrogen concentration was more relevant (61%) to the reforming temperature, hydrogen concentration in syngas and steam usage, hydrogen yield was more relevant (63%) to the dosage of activation agent and steam usage. Partial dependence plot (PDP) analysis provided the optimal range of processing parameters for maximized production of target products.