Gaps in current numerical models highlight challenges in accurately simulating photovoltaic (PV) systems, particularly their sensitivity to weather forecast outputs and the impact of environmental factors such as wind speed and humidity. These factors introduce discrepancies between simulated and real-life PV system performance. This study investigates the potential of open-source PV physical models, leveraging pvlib-python, to address these challenges. The research focuses on enhancing the accuracy of these models to generate synthetic data that closely replicate real-world conditions. Using the single diode equation implemented in pvlib-python, DC power production data from a real-life PV module is simulated, with initial discrepancies identified and mitigated through the integration of Huld model. This integration aims to improve the accuracy of synthetic data, thereby bridging the gap between simulated and actual performance metrics. The efficacy of the enhanced predictive models is evaluated using historical field measurements from a PV system testbed, demonstrating advancements in simulating PV system behaviors with enhanced accuracy and reliability. The results unveil a significant enhancement in synthetic data generation using PVLib and Huld model.

The emergence of novel technologies is significantly contributing to the study of new approaches to the development of energy Cyber-Physical systems. Technologies like IoT, Big Data, Machine Learning and AI are becoming so tightly coupled to be considered as a singular technological framework. Such a framework is at the basis of the Digital Twin technology, which allows to interact with a real-world asset by referring to its digital counterpart. Being potentially applied to any kind of physical asset, the Digital Twin approach can create a synergy for a comprehensive modeling, monitoring, and forecasting of the same asset. In the energy domain, Digital Twins can facilitate the decision making by providing interactive dashboards to the stakeholders within a metaverse instance. These dashboards can support what-if analyses, forecasts on energy demand and production, and remote control of the physical assets to optimize their performance. This paper proposes a framework for developing a Cyber-Physical solution in the energy domain. The framework consists of three building blocks called Physical Asset, Data Infrastructure, and Digital Twin. The Physical Asset and the Digital Twin blocks correspond to the typical counterparts of a Cyber-Physical system, whilst the Data Infrastructure is delegated to support their synchronization and/or interaction. To assess the proposed framework, a real testbed was developed as Cyber-Physical system to implement the digital twin of a smart photovoltaic station.

The Energy System is a critical infrastructure (CI) classified within the realm of Cyber-Physical-Social Systems due to its integration of computerized control systems and society’s consumption patterns, interfacing with the physical entities involved in production, transmission, and distribution. As we navigate cutting-edge technologies, sustainable development, and the digitization of essential societal services, new challenges emerge for energy systems. One cutting-edge technology addressing the complexity of Cyber-Physical-Social CIs is the Digital Twins (DT). The present study delves into the implementation of DT within the Energy Systems domain. Search results from three prominent digital libraries (SCOPUS, WOS, IEEE) highlight the novelty of this subject, which has gained traction within the sector since 2018. The findings of our Systematic Literature Review (SLR) reveal notable enhancements in the resilience of Smart Energy Systems through the application of DT. This review thoroughly investigates the efficacy of DT utilization, explores its achievements, and applications, and confronts the associated challenges.

Green energy production is typically decentralized, and the ecosystem of production, transmission, and distribution differs significantly from centralized systems. Therefore, ensuring the resilience of green energy infrastructure demands a distinct approach, particularly regarding the security and safety aspects of these CIs. Green Energy CIs have less inherent protection, along with ancillary protection facilities compared to conventional power plants. This underscores the need to leverage AI to enhance the safety and security of green energy infrastructures, providing efficient and cost-effective solutions. This study aims to provide a comprehensive overview of AI implementations for enhancing the security and safety of green energy. Although this study is a work in progress, the present article will specifically delve into the resilience aspects of green energy infrastructures. Given the focus on AI implementation and data-driven solutions, we approach energy systems from a cyber-physical and societal perspective, emphasizing their broader impact on society. The ongoing study has unveiled significant improvements in resilience through the application of AI methods and data-driven models, such as machine learning, deep learning, neural networks, multiagent systems, big data, and data mining. Furthermore, we explore the challenges associated with integrating AI into green energy systems and investigate its various applications. This exploration aims to identify key features that will guide the development of novel approaches to enhancing the resilience of green energy systems through AI-based solutions for security and safety. Finally, results show a significant gap in the safety applications of AI. It received the least attention in the articles While the term ”safety” is frequently mentioned, even when the article’s primary focus is not on safety applications.