An enhanced Adaptive Crossover-based Artificial Rabbits Optimization algorithm for solving optimal power flow under renewable energy uncertainty in power systems

This paper presents an enhanced version for the Artificial Rabbit Optimization (ARO), called Adaptive Crossover-based Artificial Rabbit’s Optimization (ACARO), which builds upon the ARO framework and incorporates mechanisms designed to improve local search effectiveness and maintain population diversity. To validate the performance of ACARO, it is applied to solve the optimal power flow (OPF) problem, specifically under the uncertainties associated with renewable energy sources. The IEEE 30-bus and 57-bus systems are modified by substituting selected thermal generators with two wind turbines and one solar photovoltaic (PV) unit, respectively. The stochastic nature of renewable generation is modeled using Weibull and lognormal distributions, while reserve and penalty costs are introduced to represent deviations from forecasted output. Variability in load demand is also accounted for through standard probability density functions. Additionally, operational constraints, such as generator ramp rate limits, are integrated into the overall formulation. Performance evaluation involves using 23 benchmark test functions, with ACARO being compared against seven established optimization algorithms through rigorous statistical analysis. The results indicate that ACARO outperforms both the original ARO and other competing methods, achieving a minimum total generation cost of 781.422 $/h for the modified IEEE 30-bus system (Case 1), which is 0.07% lower than the standard ARO (781.961 $/h) and 1.92% lower than the worst-performing benchmark (SFOA: 782.918 $/h). For the IEEE 57-bus system (Case 9), ACARO reduced the total generation cost to 1,281.34 $/h, outperforming all compared algorithms while satisfying all operational constraints. These findings indicate improved performance relative to the other studied algorithms. This integration is specifically designed to overcome the premature convergence and diversity loss inherent in the original ARO framework, providing a more effective solution for complex, stochastic optimization problems such as renewable-integrated OPF.