Green hydrogen projects face significant market risk from volatile electricity costs, uncertain renewable supply, and unstable hydrogen and ammonia prices. We present a two-stage stochastic optimization framework that integrates long-term financial contracts into the joint design and operation of an integrated hydrogen–ammonia plant. The model endogenizes Hydrogen Purchase Agreements (HPAs) and Contracts-for-Difference (CfDs) as first-stage decisions together with capacities of electrolysis, Haber–Bosch synthesis, air separation, storage, and grid connection. Second-stage decisions represent hourly dispatch with detailed material and power balances under scenario-based uncertainty in renewable availability and commodity prices. The objective maximizes expected net present value penalized by Conditional Value at Risk (CVaR) to reflect downside protection and bankability.

The problem is formulated as a mixed-integer linear program. We solve the extensive form via sample average approximation (SAA) and implement an integer L-shaped method with multi-cuts. The master problem includes capacity and contract variables and a linear CVaR reformulation, while scenario subproblems are linear operational models that generate feasibility and optimality cuts. Warm starts from a risk-neutral solution and contract screening are used to improve convergence. Computational experiments report solution quality and convergence behavior as functions of scenario count and risk aversion and evaluate outcomes using expected NPV, CVaR, and probability of loss, including out-of-sample performance assessment.

A case study motivated by export-oriented projects in Chile evaluates contractual portfolios by comparing no contracts, HPA-only, CfD-only, and joint portfolios against risk-neutral and risk-averse baselines. The analysis characterizes how contracting interacts with optimal sizing and operations across regimes and discusses implications for policy programs offering bankable offtake or price-stabilization instruments in green hydrogen–ammonia value chains.

The rapid penetration of renewable energy systems has intensified operational challenges related to intermittency, decentralized architectures and data-driven decision making. In addition, artificial intelligence (AI) techniques are being increasingly adopted into renewable energy applications, as shown in the existing studies. However, these studies are often fragmented, focusing on isolated forecasting or optimization tasks without providing an integrated design perspective. This study addresses this gap by systematically reviewing and synthesizing recent AI-based approaches for forecasting, optimization and control of renewable energy systems.

A structured literature analysis was conducted covering machine learning, deep learning and reinforcement learning techniques applied to solar, wind, hydropower, energy storage, and hybrid renewable systems. Forecasting models, including Long Short-Term Memory (LSTM), Gated Recurrent Units (GRUs), Convolutional Neural Networks (CNNs) and hybrid architectures, were comparatively analyzed with respect to prediction horizon, data requirements, and operational relevance. Optimization and control strategies based on genetic algorithms, particle swarm optimization, and reinforcement learning were examined for battery scheduling, hybrid system sizing and market participation.

The analysis shows that these deep learning-based forecasting models consistently outperform old traditional statistical approaches. These modes enabled improved operational planning and higher renewable penetration. Predictive maintenance using SCADA data significantly enhances fault detection and asset reliability, while learning-based optimization techniques support adaptive and resilient energy system operation. Based on these findings, a unified AI-enabled design framework is proposed to guide the selection and integration of forecasting, optimization and control techniques for renewable energy systems.

Residential electricity consumption is an important part of the general system. Modern trends in the use of remote communications and remote work only enhance the importance of this part.

The total amount of energy consumed and the stability of consumption are important characteristics of an individual profile. In our opinion, it is the combination of these two characteristics that is significant for the tasks of forecasting consumption and ensuring system stability.
Cases when consumption has high volumes and low stability are undesirable for both the individual consumer and the overall system. However, such cases may go unnoticed by owners.

In this presentation, we introduce a model of dynamic classification of residential electricity consumption, taking into account volume and stability. The availability of modern smart-grid technologies allows for monitoring individual consumption and obtaining time series of readings. For stability modeling, the coefficient of auto-similarity is used, which is dynamically calculated based on time series of hourly spaced readings.

The model has been successfully tested on real data of Swedish residentials (Zimmermann, J.P. (2009), End-use metering campaign in 400 households in Sweden. Assessment of the potential electricity saving. Enertech). Cases with low (0.20) and high (0.80) readings of the coefficient of auto-similarity were considered.

In our opinion, the proposed model has good prospects for implementation in practical devices for individual consumption monitoring. This can have a positive effect on reducing the cost for the individual consumer, as well as contributing to the stability of the overall system.

This paper proposes the design of an optimal power management and dispatch strategy among networked microgrids to enhance electrical grid resilience after disconnection. The main contribution of this paper is the design of an optimal strategy to ensure power system stability in the event of generator disconnection. This is achieved through optimal power sharing among multiple microgrids connected at the load bus, combined with smart power dispatch of renewable energy sources. The proposed strategy includes three coordinated stages involving optimal power management of renewable sources in each connected microgrid, optimal sharing of interconnected microgrids, and intelligent load shedding. Each microgrid includes a solar PV generator, concentrating solar power (CSP), and a wind farm. In addition, a multi-energy storage system was installed in each microgrid, comprising Redox Flow Batteries (RFBs), Superconducting Magnetic Energy Storage (SMES) and Fuel Cells (FCs). Moreover, an optimal Load Frequency Control (LFC) loop was applied to cope with system load disturbances or climatic changes. The main objective was to design a decentralized controller using nature-inspired optimization algorithms. For this aim, an optimal Fuzzy-PIDN controller was designed using a recently optimized algorithm called the Crayfish Optimization Algorithm (COA) , which was used to find the best controller parameters to enhance the power management of each green source, the frequency control using multiple storage units, and optimal microgrid power sharing to support the main grid frequency stability and control in case of load disturbances. Several case studies have been conducted to prove the validity of the proposed strategy. It can be concluded from the obtained results that multi-stage optimal power management and control ensure optimal sharing between interconnected multi-microgrids to increase resilience and energy autonomy after grid disconnection, and can also improve dynamic power system stability.

Accurate solar irradiance forecasting is essential for the secure integration of photovoltaic energy into power grids, especially in tropical regions with high intermittency [1]. While deep learning models like Long Short-Term Memory (LSTM) networks have been applied to this task, the potential of enhanced architectures, specifically Bidirectional LSTM (BiLSTM) and models incorporating attention mechanisms, for univariate hourly forecasting in tropical climates remains a significant research gap. This study addresses this gap by investigating whether these advanced AI architectures improve prediction accuracy in the context of Northeastern Brazil. We developed and rigorously compared four deep recurrent neural network models: LSTM, LSTM with attention, BiLSTM, and BiLSTM with attention. The models were trained on hourly data (2022–2024, 7:00–22:00 UTC) from João Pessoa, PB, preprocessed with Min-Max normalization [0.1, 0.9], and evaluated using standard metrics (RMSE, MAE, and R²) [2]. Performance was assessed robustly through 10 executions per model with 12- and 24-hour lookback windows. Contrary to expectations from recent AI trends, the results demonstrate that while BiLSTM architectures consistently outperformed standard LSTM, the integration of attention mechanisms degraded forecast performance across all configurations. This key finding suggests that for univariate solar irradiance series in tropical Brazil, the added complexity of attention gates may not capture beneficial additional temporal dependencies. This study provides empirical evidence for optimal model selection (BiLSTM) in regional grid integration applications and challenges the assumed universal utility of attention mechanisms in time-series forecasting for renewable energy.

References:

[1] Ministério de Minas e Energia. Plano Nacional de Energia 2050. Empresa de Pesquisa Energética: Brasília, Brazil, 2020.
[2] de O. Santos, D.S., Jr.; et al. Solar Irradiance Forecasting Using Dynamic Ensemble Selection. Appl. Sci. 2022, 12 (7), 3510.

In recent years, the transformation of micro-grid laboratories has accelerated toward more integrated, real-time, and data-driven environments. The evolution from early control-validation setups to advanced Power Hardware-in-the-Loop (PHIL), Co-Simulation, and Digital Twin (DT) frameworks has enabled a new generation of smart-grid experimentation emphasizing interoperability, scalability, and cyber-resilience. Modern testbeds now merge real-time simulators, HELICS/FMI-based middleware, and cloud-linked monitoring systems to create hybrid validation environments that reproduce complex grid interactions with high fidelity and automation. This convergence allows simultaneous testing of control, communication, and operational layers while bridging laboratory models with live field data. However, challenges remain in affordability, latency control, data standardization, and reproducibility, particularly for universities and utilities in resource-limited regions. This study presents a systematic review of micro-grid testbed developments between 2015 and 2025, classifying and comparing representative PHIL, Co-Simulation, and Digital Twin architectures. This study highlights open-architecture design approaches, interoperable standards, such as IEC 61850, IEEE 2030.5, and MQTT/REST, and cybersecurity-ready interfaces suitable for distributed validation. A five-step implementation roadmap is proposed to guide the development of low-cost, reproducible, and industry-aligned laboratories. The findings demonstrate that the convergence of real-time simulation, cross-domain synchronization, and Digital Twin intelligence represents a decisive step toward the next generation of sustainable, validated, and data-centric micro-grid testbeds.

Induction motors constitute the operational backbone of contemporary industrial systems, and their unexpected failure can result in severe downtime, significant productivity loss, and substantial maintenance expenditure. Conventional maintenance approaches—primarily corrective and preventive—often fail to identify early-stage degradation, making them inadequate for modern industrial demands. To address this gap, the present study proposes an AI/ML-based predictive maintenance framework capable of real-time monitoring and automated fault diagnosis in three-phase induction motors. Multimodal sensor data, including vibration, temperature, current, and voltage signals, were acquired using an ESP32-based IoT architecture and subsequently analyzed through machine learning algorithms. Feature engineering incorporated Fast Fourier Transform (FFT) and statistical indicators such as Root Mean Square (RMS), kurtosis, and crest factor to enhance the quality of the diagnostic features. A Support Vector Machine (SVM) classifier was developed and achieved an accuracy of 95–98%, with a precision of 92%, recall of 95%, and an AUC of 0.97 for classifying faults such as bearing damage, rotor imbalance, and electrical irregularities. Additionally, MATLAB/Simulink-based vector control simulations validated the dynamic behavior of the motor under varying load and fault conditions. Overall, the integration of AI/ML with IoT demonstrates significant improvements in reliability, reducing unscheduled downtime by up to 25% and enhancing energy efficiency by 8–12%, thereby offering a scalable, Industry-4.0-ready solution for intelligent predictive maintenance.

Structural Health Monitoring (SHM) has increasingly shifted toward smart-material-based solutions capable of providing continuous and real-time assessment of structural integrity. Among these approaches, the Electro-Mechanical Impedance (EMI) technique has emerged as a highly sensitive method for detecting incipient damage through the electromechanical coupling characteristics of piezoelectric (PZT) sensors. This study contributes to EMI-based SHM by numerically investigating the combined effects of structural damage and temperature variations on the impedance response of a PZT patch bonded to an aluminum beam.

The methodology relies on three-dimensional finite element modelling performed in ANSYS Multiphysics, incorporating temperature-dependent material properties for both the host structure and the PZT sensor. Several damage scenarios were simulated by introducing cracks of varying depths at a fixed distance from the sensor. Temperature was varied from 25°C to 85°C to evaluate its influence on the electrical impedance signature. Harmonic analyses were conducted over a high-frequency range (18.5–21 kHz) to obtain both the real and imaginary components of the impedance. Damage indices such as Root Mean Square Deviation (RMSD) and Correlation Coefficient Deviation Metric (CCDM) were computed to quantify structural changes.

The results show that increasing temperature produces notable horizontal and vertical shifts in impedance peaks, while increasing crack depth leads to frequency shifts and the emergence of new resonance peaks due to stiffness reduction. Although temperature effects can mask true damage signatures, the application of a cross-correlation-based compensation technique successfully mitigates these distortions, enabling clear distinction between temperature-induced variations and actual structural deterioration.

Overall, the numerical findings confirm that the EMI technique, when combined with adequate temperature compensation strategies ,offers high sensitivity and robustness for reliable, continuous structural health monitoring.

Short-term wind information is important for operating small wind plants and for planning other distributed energy resources. In this study, we work with one year of measurements from the Termas station and build a simple, reproducible workflow that goes from raw data to ten-minute wind speed forecasts. The original file contains 193,493 records with non-uniform timestamps and some suspect values near zero. We first standardise the timestamps, compute 10-minute averages and select only complete days with 144 samples, obtaining 52 days and 7,488 valid points. Days with extremely low daily means are removed and very small speeds are corrected to reduce obvious sensor and logging errors.

The cleaned series is then described with daily and global Weibull distributions. For the global fit, we obtain a shape parameter k = 1.27 and a scale parameter c = 1.40 m/s, which point to a low to moderate wind regime with frequent calm periods. For the forecasting step, we form supervised samples using 24-hour sliding windows (144 time steps) to predict the next 10-minute value. A single-layer SimpleRNN network is trained using a chronological split into training, validation, test and a final 10% hold-out set.

On the hold-out set, the network reaches a mean absolute error of 0.28 m/s and an R² of 0.78. These results suggest that even a modest recurrent model can follow the short-term variability of the wind speed at this site and that the proposed workflow can be reused with longer records or other stations.

Fixed-bottom or floating offshore structures are the foundations, platforms, and associated infrastructure that support oil and gas production systems, offshore wind turbines, and cabling. The remote nature of these structures and the harsh environment with high variability in wind, waves, currents, and weather make construction activity very difficult and unpredictable; variation in the schedule can lead to high construction vessel and personnel costs. The adoption of artificial intelligence using trends observed in historical data can help achieve more accurate construction cost and schedule predictions, reducing the capital expenditure cost of installation.

A resource-significant activity, sometimes called a resource-critical activity or high-resource-demand activity, is an activity on a construction or project schedule that consumes a disproportionately large share of one or more resources compared with others. Plant Design Modelling (PDM) is a digital process that creates and manages a detailed 3D-model of a building's physical and functional characteristics and semantic information, such as cost and schedule. PDM serves as a single source of truth for multidisciplinary activities and, therefore, serves as a rich data source for various construction applications, including project scheduling and cost estimation. Neural networks (NNs), a subset of machine learning algorithms inspired by the human brain, excel at identifying patterns in complex datasets and making predictions, such as forecasting costs based on nonlinear relationships and historical trends.

We used data extracted from Aveva’s-EverythingPDM of an offshore structure modification project, focusing on installation activities to create a dataset for machine learning model training. The extracted information includes geometry, material types, quantities, and spatial relationships between elements. The structured data extracted exhibit nonlinear patterns; therefore, they are analysed using ordinary and regularised linear regression models, as well as neural networks (NNs). NN models show a superior ability to predict the nonlinear nature of offshore construction activities' time.