One of the main barriers for the wide penetration of fuel cell electric vehicles is the lack of proper infrastructures for hydrogen transportation that hinders the implantation of refuelling stations. This barrier could be overcome by deploying onsite hydrogen generators based on mature electrolysis and hydrogen storage technologies. This way, the necessity of hydrogen transportation is avoided. In addition, electrolysers can be onsite supplied by means of renewable generators like photovoltaic panels, while the produced hydrogen can also be destined to generate electricity through fuel cells thus obtaining a monetary revenue. Thereby, the economy of the system may be improved in order to make viable this kind of infrastructures. However, the optimal coordination of the different assets is challenging and requires the use of energy management tools to pursue the optimal performance of the installation. In this kind of infrastructures, the energy management problem is performed under substantial uncertainties; moreover, these unknown parameters have a very different character. Thus, while energy pricing and renewable generation can be forecasted using conventional techniques, refuelling demand is highly unpredictable. To this end, this paper proposes a novel stochastic-interval model for the optimal scheduling of photovoltaic-assisted refuelling stations. The new proposal uses interval notation to model the inherent uncertainty of renewable generation and energy pricing, while the vehicle demand is modelled using a more suitable approach based on scenarios. In this regard, a comprehensive stochastic model for fuel cell electric vehicles is developed, which is based on reported driving behaviour and common characteristics of commercial vehicles. To solve the problem subjected to uncertainties, an iterative solution methodology is developed which allows adopting risk-seeker and risk-averse operational strategies. A case study is analysed to validate the new proposal and discussing the importance of the different economic activities that can be exploited in refuelling stations. Results reveal the importance of selling energy to the grid in order to complement the revenues obtained from refuelling; however, this process is highly impacted by uncertainties and the operational strategy, observing variations up to 50% in the total profit depending on the strategy adopted.

燃料电池电动汽车广泛渗透的主要障碍之一是缺乏适当的氢气运输基础设施，这阻碍了加氢站的植入。可以通过部署基于成熟电解和储氢技术的现场氢气发生器来克服这一障碍。这样就避免了氢气运输的必要性。此外，电解槽可以通过光伏板等可再生发电机进行现场供应，而产生的氢气也可以通过燃料电池发电，从而获得货币收入。因此，可以提高系统的经济性以使这种基础设施可行。然而，不同资产的最佳协调具有挑战性，需要使用能源管理工具来追求装置的最佳性能。在这种基础设施中，能源管理问题是在很大的不确定性下进行的；此外，这些未知参数具有非常不同的特征。因此，虽然可以使用传统技术预测能源定价和可再生能源发电，但加油需求是高度不可预测的。为此，本文提出了一种新颖的随机区间模型，用于光伏辅助加油站的优化调度。新提案使用区间表示法来模拟可再生能源发电和能源定价的固有不确定性，而车辆需求则使用基于情景的更合适的方法来模拟。在这方面，基于报告的驾驶行为和商用车的共同特征，开发了燃料电池电动汽车的综合随机模型。为了解决受不确定性影响的问题，开发了一种迭代解决方案方法，该方法允许采用风险寻求者和风险厌恶者的操作策略。分析案例研究以验证新提案并讨论可在加油站开发的不同经济活动的重要性。结果揭示了向电网出售能源以补充从加油中获得的收入的重要性；然而，这一过程受到不确定性和运营策略的高度影响，根据所采用的策略，总利润的变化高达 50%。