<![CDATA[Production scheduling in modern manufacturing must simultaneously address operational efficiency, environmental sustainability, and customer satisfaction. As industries face increasing pressure to reduce carbon emissions while maintaining competitiveness, traditional single-objective scheduling approaches prove insufficient. Non-identical parallel machine scheduling problem (NIPMSP), where machines possess heterogeneous processing capabilities and environmental impacts, represents a prevalent configuration in manufacturing facilities. However, existing research has not simultaneously optimized makespan, carbon emissions, and tardiness—three critical objectives reflecting productivity, sustainability, and service reliability.In this study, we developed a comprehensive mathematical model with 18 constraints and proposed Multi-Objective Adaptive Large Neighborhood Search (MOALNS) incorporating five destroy operators, four repair operators, and adaptive weight mechanisms. Performance evaluation against NSGA-II across eight problem instances using five metrics (Hypervolume, Inverted Generational Distance, Diversification Metric, Spacing Metric, CPU Time) with rigorous statistical testing (Wilcoxon signed-rank test, Cohen's d) revealed that MOALNS significantly outperforms NSGA-II in solution diversity (p < 0.05 in 50% of instances, d = 6.31) and spacing uniformity (75% win rate, d = 0.70), with comparable Pareto front coverage. This superiority comes at computational cost (192% slower), acceptable for offline planning but not real-time applications. This research provides evidence-based algorithm selection guidelines for sustainable manufacturing, demonstrating that environmental objectives can be integrated without sacrificing operational efficiency or delivery reliability.]]>
