Real-world Optimization using Adaptive Fusion Optimization: energy-efficient scheduling and network optimization analysis

Abstract

Introduction: This study investigates the effectiveness of Adaptive Fusion Optimization (AFO) for solving energy-efficient economic load dispatch problems on standard IEEE benchmark power systems using the MATPOWER environment under a fixed-budget, repeated-run protocol. Objective: The objective is to evaluate the performance, stability, feasibility, and scalability of AFO and compare it fairly with GA, PSO, DE, and ACO using identical evaluation budgets and population settings. Method: Experiments are conducted on IEEE 30-, 57-, and 118-bus systems. Each algorithm is executed for 30 independent runs. Performance is assessed based on mean generation cost, standard deviation, feasibility rate, and statistical validation using Friedman and Wilcoxon tests. Sensitivity analysis is performed under ±5% and ±10% load variations with ramp-rate and operational constraints enabled. Results: On the IEEE 30-bus system, AFO achieves a mean cost of 801.12 $/h with a 0.73 $/h standard deviation and 100% feasibility. On the IEEE 57-bus system, AFO maintains 96%, feasibility while delivering the lowest mean cost with reduced variability under load and ramp constraints. On the IEEE 118-bus system, AFO achieves a mean cost of 128,740.5 $/h with 100% feasibility, demonstrating strong scalability. Statistical tests confirm the significance of these improvements, and sensitivity analysis shows near-proportional cost scaling with low sensitivity to load perturbations. Conclusions: AFO provides stable, feasible, and scalable performance for economic load dispatch, outperforming traditional metaheuristics under identical conditions while maintaining robustness to demand variations through effective constraint handling.