Abstract: Hydrogen aroused a great interest in the last decades as a promising energy vector for the energy transition as well as renewable energy storage. This article investigates the technical feasibility of a hydrogen-powered fleet, in the context of a microgrid with renewable generation. The hydrogen demand is satisfied through the electrolyser powered by local RES and periodical external refueling of a local hydrogen storage system. The model of the microgrid is implemented in the open-source GNU Octave environment and is exploited to size the local storage capacity. Different types of vehicles were considered and hydrogen pressure levels were distinguished among the system components. The model was applied to the waste collection fleet and the headquarters of ASM S.p.A., a utility located in the center of Italy, chosen as a case study. The dataset of 143 vehicles was used to model the hydrogen-powered fleet. Three scenarios were evaluated: hydrogen production using i) Reverse Power Flow, ii) PV production and iii) electricity drown from the grid. The storage capacity depends on the scenarios and in the case study it ranges from about 35 to 47 m3, in comparison with 13.6 m3 of current diesel refueling system, assuming the same frequency of external refueling.

Abstract: Artificial Intelligence (AI) is expected to radically reshape the energy sector value chain, by improving business processes performance, while increasing environmental sustainability and propagating high social value among citizens. Additionally, a vast amount of energy data is available, coming from several sources including smart grid sensors, simulators and open data sources. However, the energy sector is characterised by uncertain business cases, fragmented regulations, immaturity of standards and lack of high-end ICT workforce of EPES stakeholders. Additionally, the lack of interoperability across data stream providers, of energy data ownership and sharing, and of a holistic, safe and cooperative AI perspective amongst the EPES community actively hinder the effective integration of AI services in the sector. This publication presents the I-NERGY modular framework for supporting AIon-Demand in the energy sector by capitalising on state-of-the-art AI, including resources of the AI4EU platform, as well as IoT, data analytics, adaptive learning and digital twin technologies. The solution, developed within the EU funded I-NERGY project, will enable AIbased cross-sector multistakeholder analytics tools for integrated and optimised smart energy systems management, based on interoperable data exchange. The project will also manage Open Calls for the delivery of innovative AI-driven energy services through the provisioning of financial support to third parties.

Abstract: Artificial Intelligence (AI) is expected to radically reshape the energy sector value chain, by Abstract: Artificial Intelligence (AI) holds the premise to transform the energy sector and the underlying value chain; scarcity of AI expertise in the energy community, fuzzy and unclear regulations on access to data, standards’ immaturity and uncertain business cases are hampering though the full exploitation of its potential. In this context, the goal of this paper is to present the I-NERGY project, an Innovation Action that targets to promote AI in the energy sector by reinforcing the AI-on-demand (AI4EU) platform service offering and ecosystem. To this end, the paper introduces the I-NERGY project concept, the domain challenges it addresses and the target audience towards which it is addressed, exposes the project technical solution and pilot use cases that respectively incarnate, and exemplify and validate it and emphasizes its open call mechanism for providing financing support to third SMEs for energy use cases and AI services proliferation.

Abstract: Modern power systems introduce new dynamics, which may require the appropriate selection of the modeling method for each type of dynamic simulation. In this paper, electromagnetic transient (EMT), Root Mean Square (RMS), and Shifted Frequency Analysis (SFA) modeling are compared in the scope of the dynamic simulation of transient stability analysis. The comparison is carried out by analyzing the accuracy of the simulation results as a function of the simulation step size. The evaluation is conducted for classic and low inertia systems for the accuracy of rotor angle transients and the critical clearing time, being the variables of interest in transient stability analysis. The real-time open source simulator DPsim is employed. The main advantage presented by DPsim, is the possibility to run the same simulation scenario with the same solver, but in three different modeling domains. This powerful feature enables a systematic comparison between the modeling methods. With respect to transient stability analysis, the results of the comparative analysis support the usage of SFA, both for classic and low inertia systems. SFA incorporate more dynamics than RMS simulation into the models, while at the same time allowing larger step sizes than EMT simulation.

Abstract: In power grids, short-term load forecasting (STLF) is crucial as it contributes to the optimization of their reliability, emissions, and costs, while it enables the participation of energy companies in the energy market. STLF is a challenging task, due to the complex demand of active and reactive power from multiple types of electrical loads and their dependence on numerous exogenous variables. Amongst them, special circumstances—such as the COVID-19 pandemic—can often be the reason behind distribution shifts of load series. This work conducts a comparative study of Deep Learning (DL) architectures—namely Neural Basis Expansion Analysis Time Series Forecasting (N-BEATS), Long Short-Term Memory (LSTM), and Temporal Convolutional Networks (TCN)—with respect to forecasting accuracy and training sustainability, meanwhile examining their out-of-distribution generalization capabilities during the COVID-19 pandemic era. A Pattern Sequence Forecasting (PSF) model is used as baseline. The case study focuses on day-ahead forecasts for the Portuguese national 15-minute resolution net load time series. The results can be leveraged by energy companies and network operators (i) to reinforce their forecasting toolkit with state-of-the-art DL models; (ii) to become aware of the serious consequences of crisis events on model performance; (iii) as a high-level model evaluation, deployment, and sustainability guide within a smart grid context.

Abstract: Internet of things (IoT) along with big data technologies can accrue significant added value in several domains and improve people’s everyday life. One of the domains that can be benefitted the most by the aforementioned technologies is Smart Buildings. This is because, several aspects of people’s everyday lives can be improved through IoT services, such as energy consumption, health, heating, building security and more. IoT services can be divided to near real-time, and static based on the time that they require in order to return results. Significant amount of research papers has been dedicated to the second for services such as energy forecasting, while for near real-time services there are not so many publications, while, most of the existing ones focusing mostly on obtaining meaningful results. In this publication we propose a conceptual architecture for building a near real-time Anomaly Detection service for smart buildings using the Fog Computing paradigm, to achieve scalability and low latency. Moreover, we provide a technical glance of the proposed solution, suggesting specific technologies for each functionality as well as restrictions for each technology. It is worth mentioning that the proposed approach can be easily adapted for other near real-time services with little modifications.

Abstract: A systematic and systemic analysis of historical data in power systems can contribute to the creation of condition based monitoring solutions for critical assets as circuit breakers. This work presents a methodology that automatic processes event-based data in COMTRADE format to obtain relevant metrics used in asset management. It considers the data processing, fault detection, classification and analysis stages at both, device and system level to aggregate and provide relevant metrics to end users. The methodology is validated using a sub-set of real life COMTRADE files from faults that occurred in the Portuguese Transmission System, between the years 2011 and 2021. The outcomes of this validation step are herein presented as well.

Abstract: The transition to Smart Grids increases the complexity of power grids by involving many more interdependent actors and integrating additional information and communications technology. To provide a common basis for Smart Grid data representation and exchange, the standardized Common Information Model (CIM) has been introduced and extended, i. a., by the Common Grid Model Exchange Specification (CGMES). An increasing acceptance by power grid operators and other actors has made CIM and CGMES more and more relevant. However, the implementation of CIM / CGMES support in software projects appears to be challenging due to the complexity of CIM / CGMES and the ongoing standardisation process with iterative adaptations. Thus, the main contribution of this paper is the presentation of a methodology for an automated generation of programming language specific code from CIM / CGMES specifications. The approach is based on the use of a template language and enables to keep software projects fully compliant with CIM / CGMES specifications. The paper outlines the process of code generation and the consecutive codebase integration for a JavaScript based CIM / CGMES web editor and for two CIM / CGMES de-/serialiser libraries in C++ and Python. The approach is evaluated in use cases involving the visualisation and simulation of a benchmark grid.

Abstract: Artificial Intelligence is increasing its relevance in the energy sector, which is under pressure to meet the growing demand for renewable energy. AI-based analytics for smart energy management can help ensure a better balance between supply and demand. In this context, the EU-funded I-NERGY project is working to evolve, scale up and demonstrate innovative AI as a service energy analytics applications and digital twin services. These are validated along pilots that span over the full energy value chain, ranging from optimized management of grid and non-grid renewable energy systems assets to improved efficiency and reliability of electricity networks’ operation. The goal of this paper is to present the innovative action that targets to promote AI in the energy sector by reinforcing the AI-on-demand (AIoD, formerly AI4EU) platform service offering and ecosystem. To this end, the paper introduces the target audience and the addressed domain challenges, exposes technical solutions and pilot use cases that exemplify and validate those solutions, and emphasizes its open call mechanism for providing financing support to third SMEs for energy use cases and AI services proliferation.

Abstract: Today’s rapidly changing marketplaces are constantly bringing new ways of transforming business operations and require companies to be flexible and dynamic. Toward this direction the use of Artificial Intelligence (AI) tools and techniques has the potential to bring significant value by transforming business operations and increasing their efficiency. Companies are already benefiting from AI solutions such as prescriptive analytics to enhance customers’ experience or achieve optimal use of limited resources. This paper analyses the opportunities and challenges deriving in the context of introducing AI solutions in Small and Medium Enterprises (SMEs) in different sectors of activity (matchmaking, earth observation, energy services, planning services, cyber physical services, digital innovation hubs) by reviewing the research design, methodology and expected results of six Horizon 2020 European projects: AI4Copernicus, AIPlan4EU, BonsAPPs, DIH4AI, I-NERGY and StairwAI. The overarching goal of these projects is the enhancement of the European AI-on-Demand (AIoD) platform by mobilising the European AI community to support businesses and sectors in accessing expertise, knowledge, algorithms and tools for successfully applying AI and thereby generating market impact.

Abstract: Net load forecasting has become increasingly complex due to the high penetration of renewables and evolving data subject to so-called concept drift, i.e., sudden and large changes in the energy flow pattern. Accurate net load forecasting is essential to prevent unexpected imbalances across all voltage levels of the electricity grid, as well as to promote the stability and reliability of the power system. Therefore, the use of accurate forecasting techniques is essential to manage and optimize the use of available resources at the TSO-DSO interface. This paper evaluates several forecasting methods, including an adaptive random forest method based on incremental learning incorporating a drift detector, a method based on a recurrent neural network using long short-term memory (LSTM), and a method based on an ensemble of models including decision trees (DT), support vector machines (SVM), extreme gradient boosting algorithms (XGBoost), and Lasso regressions. The experiment was conducted using the net load data collected at the TSO-DSO interface in Portugal, where concept drift can be observed, possibly due to increasing integration of distributed energy resources behind the meter. The study examined two scenarios. In the first scenario, the models were trained using a large training set that included significant drifts, while in the second scenario, the models were trained prior to the occurrence of the drifts. The results showed that the approach using the adaptive model is more robust to the concept drift and performs better compared to the other traditional methods, especially in scenarios where there are significant changes in the net load patterns over time.

Abstract: Short-term load forecasting (STLF) is crucial for the daily operation of power grids. However, the non-linearity, non-stationarity, and randomness characterizing electricity demand time series renders STLF a challenging task. Various forecasting approaches have been proposed for improving STLF, including neural network (NN) models which are trained using data from multiple electricity demand series that may not necessarily include the target series. In the present study, we investigate the performance of a special case of STLF, namely transfer learning (TL), by considering a set of 27 time series that represent the national day-ahead electricity demand of indicative European countries. We employ a popular and easy-to-implement feed-forward NN model and perform a clustering analysis to identify similar patterns among the load series and enhance TL. In this context, two different TL approaches, with and without the clustering step, are compiled and compared against each other as well as a typical NN training setup. Our results demonstrate that TL can outperform the conventional approach, especially when clustering techniques are considered.