Artificial intelligence-powered study of a waste-to-energy system through optimization by regression-centered machine learning algorithms

The urgent need to address the global energy crisis and manage plastic waste has prompted innovative approaches at the intersection of environmental sustainability and technology. In this study, the synergistic integration of plastic waste gasification and solid oxide fuel cells (SOFCs) as a viable solution is explored. This research leverages regression machine learning algorithms to analyze and optimize this integrated energy system. The findings indicate a notable enhancement in accuracy with the implementation of non-linear machine learning algorithms compared to linear and quadratic counterparts. Specifically, the R-squared value for the electrical machine learning algorithm saw a noteworthy rise from 0.9146 (linear) and 0.9427 (quadratic) to 0.9990 (non-linear), reflecting improvements of around 0.0844 and 0.0563, respectively. It was established that the most favorable conditions entailed a steam to plastic waste ratio of 1.6, an SOFC temperature of 1113 K, and a current density of 2000 A/m2. These parameters resulted in an electrical efficiency of 41.89 %, an exergy efficiency of 37.70 %, and emissions totaling 860.71 kg/MWh. This study represents a breakthrough by demonstrating that the integration of plastic waste gasification with SOFCs can achieve an unprecedented electrical efficiency of 41.89 % through machine learning optimization. By significantly enhancing energy recovery from plastic waste, this approach not only mitigates environmental pollution but also offers a competitive edge over existing waste-to-energy solutions, positioning it as a viable pathway toward sustainable energy practices that can benefit industries and communities worldwide.