AI-driven design and optimization of microwave radiation-induced pyrolysis systems using machine learning and metaheuristic algorithms

Microwave radiation-assisted pyrolysis holds significant promise for sustainable biomass valorization, yet effectively modeling and optimizing this complex radiation-driven process remains challenging due to its inherent nonlinearity and multi-dimensional trade-offs. This study addresses this critical gap by proposing an innovative hybrid framework that integrates advanced machine learning, metaheuristic multi-objective optimization, and systematic multi-criteria decision-making. The primary objective was to holistically model, optimize, and prioritize operating scenarios for microwave radiation-induced pyrolysis of palm kernel shells (PKS) to enhance bio-char yield and quality. The methodology combines four phases: comprehensive input–output analysis, predictive surrogate modeling using the COMBI algorithm, Pareto-based optimization via the multi-objective grey wolf optimizer (MOGWO), and final solution ranking using the weighted Tchebycheff method (WTM). The COMBI models demonstrated excellent predictive accuracy, with all objectives achieving an R 2 greater than 0.99, ensuring robust generalization and reliability. Optimization outcomes revealed insightful trade-offs among reaction time, sample mass, and nitrogen gas flow rate. Notably, higher calorific values (up to 31 MJ/kg) were achieved with shorter reaction times (≈39–40 min), smaller sample masses (≈21–23 g), and lower nitrogen flows (≈100 mL/min), whereas maximizing fixed carbon content (up to 67.6 wt%) required longer reaction times (≈42–43 min), larger feedstock quantities (≈29–30 g), and higher gas flow rates (≈140–155 mL/min). Analysis of the Pareto front showed that over 55 % of optimal solutions favored shorter reaction durations, while nitrogen flow optimization consistently converged on low to medium ranges. Final ranking using WTM yielded ten distinct decision scenarios, each aligning with different technical and practical objectives. This flexibility highlights the framework's real-world applicability, offering a robust decision-support toolkit that enables practitioners to dynamically align pyrolysis operations with shifting market demands, energy efficiency goals, and environmental constraints.