Tidal stream turbines (TSTs) are crucial for renewable energy generation but face challenges from marine biofouling , significantly impacting their efficiency. Traditional methods for predicting performance and detecting biofouling rely on empirical models and manual inspections, which are often time-consuming and less accurate. This study introduces RegStack, a novel machine learning-based ensemble model, to enhance the prediction of power and thrust coefficients ( C P and C T ) and accurately classify biofouling levels in TSTs. Unlike conventional models, RegStack integrates L1 and L2 regularization into a stacking framework, improving robustness, generalization, and interpretability. The model dynamically balances the strengths of multiple regression and classification algorithms , optimizing predictive accuracy while mitigating overfitting. Comprehensive experiments were conducted using an extensive dataset of tidal stream turbine performance metrics under varying operational and environmental conditions. The RegStack model outperformed conventional approaches, achieving a coefficient of determination ( R 2 ) of 0.989 for performance predictions, with minimal mean absolute error (MAE) and mean squared error (MSE). Additionally, the model achieved 98.39% classification accuracy , with precision and recall of 0.97, and an F1-score of 0.97 in biofouling detection, demonstrating its effectiveness in real-time turbine health monitoring. By providing an automated, data-driven alternative to traditional methods, this study underscores the potential of advanced machine learning techniques in optimizing TST operations, reducing maintenance costs, and enhancing the reliability of marine renewable energy systems. The proposed RegStack model offers a scalable framework applicable to other renewable energy technologies , supporting sustainable energy advancements.

MOSFETs are used in several industrial applications, such as power electronics to supply switching loads, motor power supply, switching power supply, audio amplifier, inverter etc. The estimation of the RUL (Remaining Useful Life) of these devices is very interesting for the industry. Indeed, it allows gaining availability as well as an improvement of the safety of the system. The modern industry is focused on PHM (Prognostics and Health Management) to estimate the RUL and to design modern expert systems that ensure a good reliability to the industry.In this work, authors have developed a new methodology based on the deep learning ANN (Artificial Neural Network) in order to predict the RUL of MOSFETs. To validate the proposed Long Short-Term Memory (LSTM) algorithm, the data of the Ames PCOE (Prognostics Center of Excellence) was used. In addition, the evaluation was performed by comparing the errors between the real MOSFET RUL and the predicted MOSFET RUL. Besides, a comparison with different algorithms of prediction existing in the literature and using the same PCOE data set was done to present a solid evaluation of the proposed LSTM Recurrent Neural Network (LSTM-RNN).

This work addresses the important and yet relatively unexplored topic of degradation analysis of wave energy converters and more specifically Oscillating Water Column Turbines (OWCTs). This is achieved by developing mathematical models to simulate turbine degradation factors over time while considering environmental variability. Specifically, these factors are then combined multiplicatively to portray the overall health parameters of an OWCT system. Additionally, sensor measurements are generated based on these health parameters, encompassing various turbine-related and environmental variables such as vibration, temperature, pressure, acceleration, strain, flow, torque, rotation, wind speed, wind direction, air temperature, humidity, water temperature, solar radiation, wave height, wave frequency, wave direction, current velocity, water depth, salinity, biofouling, and seabed conditions. The generated dataset facilitates degradation analysis, enabling prognostic under uncertainty quantification. A set of adaptive time series along with small-scale learning algorithms are employed to analyze the dataset. Additionally, a wide range of metrics, and coefficient of determination, are utilized in a cross-validation scheme for generalizability evaluation. Moreover, confidence interval-based uncertainty quantification is integrated. This in-depth analysis sheds light on degradation patterns and their complexity, laying the basis for future studies into prognostics and health management of wave energy converter systems.

Prognostic learning systems (PLS) primarily rely on and delve into understanding the behavior of degradation lifecycles. Linear degradation often serves as the prevailing trend for labeling performance indicators (PI) in Remaining Useful Life (RUL) or Health Index (HI) estimations. Nevertheless, some studies explore deriving HI directly from PI, capturing state of health (SoH) variations more effectively. Disregarding the impact of operating condition disturbances on PI measurements reveals that systems do not strictly follow a linear degradation path during deterioration. At certain Health Stages (HS), a system maintains a specific SoH before continuing to deteriorate, sometimes even briefly returning to a better SoH. These multivariate conditions (MVC) arise from factors such as integrated control systems and system resilience. In this less-explored perspective, following procedural data preprocessing, this study aims to enhance data quality in terms of RUL ground truth labeling by considering aspects of temporary stability and performance backup. To assess data quality improvements, a comparative study investigates various adaptive learning systems across multiple evaluation stages and metrics, focusing on generalization as the main criterion for accuracy assessment. Recognizing the inherent non-linearity in PI degradation patterns has proven to be a pivotal element in cleansing feature spaces for PLS, consequently bolstering its generalization capability. This enhancement ultimately contributes to the development of more effective maintenance schedules.

The increasing reliance on advanced power systems in critical infrastructure highlights the need to enhance cybersecurity resilience. This study tackles this crucial issue through Exploratory Data Analysis (EDA) and Representation Learning (RL). Several real-world, complex power system datasets characterized by their large size, dynamic nature, and high imbalance are examined. EDA serves as a fundamental step to provide insights into power system event scenarios, supported by a comprehensive preprocessing strategy that includes managing missing values, reducing dimensionality, denoising, removing outliers, and addressing class imbalance. Missing values are thoughtfully handled by evaluating mean values. Dimensionality reduction involves removing insignificant features that might provide misleading information, along with integrating Principal Component Analysis (PCA). Class-specific denoising and outlier removal follow, with denoising utilizing various algorithms and a diverse range of wavelet functions to improve data quality. Robust outlier removal is achieved through iterative analyses and distance-based techniques. To address class imbalance, the Synthetic Minority Over-Sampling Technique (SMOTE) is employed, ensuring a balanced and representative dataset. For RL, deep learning models such as Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), Gated Recurrent Units (GRU), and their Multiverse Recurrent Expansions with Multiple Repeats (MV-REMR) are used in a comparative study. The results demonstrate promising and satisfactory outcomes for all models, with MV-REMR delivering the best performance. This study integrates diverse techniques into a unified EDA and RL approach, offering valuable insights into power system cybersecurity forensics, addressing current challenges, and promoting resilient and adaptive cybersecurity measures in power systems.

This study addresses the biofouling challenges in Tidal Stream Turbines (TSTs) to ensure their reliable and optimal operation. In this context, it is proposed an effective methodology employing a soft voting ensemble transfer learning-based approach for the detection and extent classification of biofouling. The proposed framework incorporates essential components such as data augmentation and pre-processing, including image resizing and data segmentation, forming a comprehensive video image-based approach. To overcome the constraint of limited data, experimental investigations were conducted, resulting in the acquisition of two datasets: one from the TST platform at Shanghai Maritime University (SMU) and the other from the tidal turbulence test facility at Lehigh University (LU). The three prominent convolutional neural network models, namely Visual Geometry Group (VGG), Residual Network (ResNet) and MobileNet, trained on these datasets, demonstrate precise detection and classification of turbine conditions, achieving an accuracy of 83% for the SMU dataset and 90% for the LU dataset. The noted disparity in accuracy for the SMU dataset is attributed to its smaller size, highlighting the significant impact of dataset scale on classification performance. This study provides valuable insight into the development of effective biofouling detection and classification strategies for TST systems. • Provide a soft voting ensemble transfer learning-based tool for real-time biofouling detection and estimation. • Data augmentation using rotation, scaling, flipping, zooming, and brightening of the cropped input images. • Data pre-processing including image resizing and data segmentation using the segment anything model.

High-speed bearings play a vital role in the functionality of modern wind turbines. Accurately estimating impending failure and predicting the remaining useful life (RUL) of bearings brings significant benefits in scheduling maintenance strategies and preventing sudden turbine shutdowns. This paper discusses a reliability enhancement prognosis framework for high-speed wind turbine bearings based on a measured representative vibration signature. A health indicator is constructed using a sliding window adaptive threshold (SWAT) to map the degradation behavior of the bearings into RUL. The Gated recurrent unit (GRU)-based deep learning algorithm shows promising predictive performance in terms of mean absolute error (MAE) and relative accuracy (RA).

This paper investigates the use of transfer learning models for detecting and classifying biofouling extent in tidal stream turbines. It provides a comprehensive exploration of a soft voting ensemble approach to deal with a multiclass image detection problem. In this context, three well-known convolutional neural network models, namely VGG16, ResNet50, and MobileNetV2, are the primary focus of the proposed study and are trained on an experimental dataset issued from the Shanghai Maritime University tidal stream turbine platform.

This paper proposes to map a machine learning path forward for tidal stream turbines biofouling detection and estimation. The proposed review covers an overview of biofouling and its impact on tidal stream turbines, current techniques for detecting and estimating biofouling, recent developments, and challenges in the field, as well as several promising prospects for biofouling detection and estimation.

International audience

International audience

One of the main data-driven challenges when assessing bearing health is that training and test samples must be drawn from the same probability distribution. Indeed, it is difficult and almost rare to witness such a phenomenon in practical applications due to the constantly changing working conditions of rotating machines. In addition, collecting sufficient deterioration samples from the bearing life cycle is not possible due to the huge memory requirements and processing costs. As a result, accelerated life tests are believed to be the primary alternatives to such a situation. However, and unfortunately, the recorded samples always are subject to lack of real patterns. Therefore, in this paper, a transfer learning approach is performed to solve such kind of problem where PRONOSTICO dataset is used to assess the current procedures.

Condition Monitoring of photovoltaic systems plays an important role in maintenance interventions due to its ability to solve problems of loss of energy production revenue. Nowadays, machine learning-based failure diagnosis is becoming increasingly growing as an alternative to various difficult physical-based interpretations and the main pile foundation for condition monitoring. As a result, several methods with different learning paradigms (e.g. deep learning, transfer learning, reinforcement learning, ensemble learning, etc.) have been used to address different condition monitoring issues. Therefore, the aim of this paper is at least, to shed light on the most relevant work that has been done so far in the field of photovoltaic systems machine learning-based condition monitoring.