Enhancing stochastic planning in autonomous hybrid energy systems through an advanced arithmetic optimization algorithm and K-means data clustering

초록

In this paper, a stochastic framework for optimal sizing of a stand-alone hybrid energy system, including photovoltaic and wind turbine resources integrated with battery storage, is presented to meet an annual load while incorporating the uncertainties of resource production and load for 20 years of the project life. Monte Carlo simulation and K-means data clustering have been employed for uncertainty modeling and scenario reduction. The decision variables, such as system component sizes, are optimized using an improved arithmetic optimization algorithm (IAOA) to minimize the net present cost (NPC) and consider the reliability constraint as the probability of load supply inability (PLSI). The IAOA enhances the conventional arithmetic optimization algorithm (AOA) using Rosenbrock's direct rotational method to overcome premature convergence. Three configurations of the HES are evaluated: (Case I) a photovoltaic-battery system, (Case II) a wind turbine-battery system, and (Case III) a hybrid photovoltaic-wind turbine-battery system. Results show that Case III, with contributions from all renewable units and reserve energy management, provides the load with lower NPC and higher reliability than Case I and Case II. Additionally, the IAOA delivers the most effective solution, yielding the lowest NPC and PLSI compared to the conventional AOA. The stochastic sizing results indicate that incorporating uncertainty increases NPC and weakens reliability. Specifically, in stochastic sizing for Case III, the PLSI, cost of energy, and NPC were higher than the deterministic approach by 3.96 %, 5.17 %, and 9.15 %, respectively. The results demonstrate that the proposed stochastic sizing framework enhances the decision-making process for energy operators under uncertain conditions, providing insight into the system's costs and reliability.

키워드