This paper evaluates the performance and suitability of four different metaheuristic algorithms for optimal sizing of standalone microgrids in remote area. The studied metaheuristic algorithms are particle swarm optimization, differential evolution, water cycle algorithm and grey wolf optimization. These algorithms are applied to optimize the capacity of diesel generator, fuel tank, solar photovoltaic, wind turbine, and battery energy storage in four different AC-coupled standalone microgrids for a remote area community in South Australia. The objective function is selected as the net present value of electricity over a 20-year lifetime. The optimisation study is conducted based on the real data of annual load consumption, ambient temperature, solar insolation, and wind speed of the site. Capital, replacement, and maintenance costs of components in Australian market are incorporated for the economic analysis. An operating power reserve is maintained based on the static and dynamic reserve concepts. Uncertainty analysis based on 10-year real data of renewable energies and load consumption is conducted. Sensitivity analysis is provided for variations of the battery price and capacity. The performance of the applied algorithms is evaluated by comparing the economic and operational results, as well as the computational time and optimization convergence. It is found that differential evolution algorithm is unreliable for optimal sizing problem of the studied standalone microgrids..

本文评估了四种不同元启发式算法的性能和适用性，以优化偏远地区独立微电网的规模。研究的元启发式算法有粒子群优化、差分进化、水循环算法和灰狼优化。这些算法用于优化南澳大利亚偏远地区社区的四个不同交流耦合独立微电网中的柴油发电机、燃料箱、太阳能光伏、风力涡轮机和电池储能的容量。目标函数被选为 20 年生命周期内电力的净现值。优化研究是根据场地的年负荷消耗、环境温度、太阳日照和风速的真实数据进行的。资本，置换，和澳大利亚市场组件的维护成本被纳入经济分析。基于静态和动态储备概念维持运行动力储备。基于10年可再生能源和负荷消耗的真实数据进行了不确定性分析。针对电池价格和容量的变化提供了敏感性分析。通过比较经济和运行结果，以及计算时间和优化收敛来评估应用算法的性能。发现差分进化算法对于所研究的独立微电网的优化规模问题是不可靠的。基于10年可再生能源和负荷消耗的真实数据进行了不确定性分析。针对电池价格和容量的变化提供了敏感性分析。通过比较经济和运行结果，以及计算时间和优化收敛来评估应用算法的性能。发现差分进化算法对于所研究的独立微电网的优化规模问题是不可靠的。基于10年可再生能源和负荷消耗的真实数据进行了不确定性分析。针对电池价格和容量的变化提供了敏感性分析。通过比较经济和运行结果，以及计算时间和优化收敛来评估应用算法的性能。发现差分进化算法对于所研究的独立微电网的优化规模问题是不可靠的。