| Abstract |
Gasification stands as a transformative thermochemical process, ingeniously converting carbon-rich substances like methane (CH4) and a spectrum of hydrocarbons, including ethylene (C2Hn), into a versatile synthesis gas (syngas). This dynamic blend predominantly comprises hydrogen (H2) and carbon monoxide (CO), presenting a potent feedstock for diverse industrial applications. In recent years, the focus on sustainable energy has intensified due to concerns about climate change, energy security, and dwindling fossil fuel reserves. Biomass energy has emerged as a promising alternative, offering the potential for a global circular economy and carbon neutrality, thanks to its abundant resources and reliable energy production. This article introduces two hybrid models that combine Least Square Support Vector Regression (LSSVR) with Dwarf Mongoose Optimization (DMO) and the Improved Grey Wolf Optimization Algorithm (IGWO). These models utilize nearby biomass data to predict the elemental compositions of CH4 and C2Hn. The assessment of both individual and hybrid models has demonstrated that integrating LSSVR with these optimizers significantly improves the accuracy of CH4 and C2Hn predictions. According to the findings, the LSDM model emerges as the top performer for predicting both CH4 and C2Hn, achieving impressive R2 values of 0.988 and 0.985, respectively. Moreover, the minimal RMSE values of 0.367 and 0.184 for CH4 and C2Hn predictions respectively affirm the precision of the LSDM model, rendering it a suitable option for practical real-world applications. Accurate predictions enable the design of systems that efficiently convert a wide range of feedstocks into valuable syngas, which can be employed to generate heat, electricity, fuels, and chemicals. By understanding and optimizing gasification processes, it becomes possible to minimize emissions of pollutants, reduce waste, and mitigate greenhouse gas emissions through carbon capture and utilization technologies. |