Search

Reliability and security of power distribution and data traffic in smart grid (SG) are very important for industrial control systems (ICS). Indeed, SG cyber-physical connectivity is subject to several vulnerabilities that can damage or disrupt its process immunity via cyberthreats. Today's ICSs are experiencing highly complex data change and dynamism, increasing the complexity of detecting and mitigating cyberattacks. Subsequently, and since Machine Learning (ML) is widely studied in cybersecurity, the objectives of this paper are twofold. First, for algorithmic simplicity, a small-scale ML algorithm that attempts to reduce computational costs is proposed. The algorithm adopts a neural network with an augmented hidden layer (NAHL) to easily and efficiently accomplish the learning procedures. Second, to solve the data complexity problem regarding rapid change and dynamism, a label autoencoding approach is introduced for Embedding Labels in the NAHL (EL-NAHL) architecture to take advantage of labels propagation when separating data scatters. Furthermore, to provide a more realistic analysis by addressing real-world threat scenarios, a dataset of an electric traction substation used in the high-speed rail industry is adopted in this work. Compared to some existing algorithms and other previous works, the achieved results show that the proposed EL-NAHL architecture is effective even under massive dynamically changed and imbalanced data. [ABSTRACT FROM AUTHOR]

In the last years, predictive maintenance has gained a central position in condition-based maintenance tasks planning. Machine learning approaches have been very successful in simplifying the construction of prognostic models for health assessment based on available historical labeled data issued from similar systems or specific physical models. However, if the collected samples suffer from lack of labels (small labeled dataset or not enough samples), the process of generalization of the learning model on the dataset as well as on the newly arrived samples (application) can be very difficult. In an attempt to overcome such drawbacks, a new deep supervised learning approach is introduced in this paper. The proposed approach aims at extracting and learning important patterns even from a small amount of data in order to produce more general health estimator. The algorithm is trained online based on local receptive field theories of extreme learning machines using data issued from a propulsion system simulator. Compared to extreme learning machine variants, the new algorithm shows a higher level of accuracy in terms of approximation and generalization under several training paradigms. • ELM-based DBN training for both unsupervised learning and supervised fine tuning stages. • Integration of locally connected combinatorial neural sub-networks into hidden nodes based on ELM-LRF. • Introduction of a regularized online OS-ELM-based learning to address adaptive training to prevent from structural risks. • Convolutional mapping and deep features reconstruction with autoencoders stack in multilayer neural network single framework. [ABSTRACT FROM AUTHOR]

Remaining Useful Life (RUL) prediction for aircraft engines based on the available run-to-failure measurements of similar systems becomes more prevalent in Prognostic Health Management (PHM) thanks to the new advanced methods of estimation. However, feature extraction and RUL prediction are challenging tasks, especially for data-driven prognostics. The key issue is how to design a suitable feature extractor that is able to give a raw of time-varying sensors measurements more meaningful representation to enhance prediction accuracy with low computational costs. In this paper, a new Denoising Online Sequential Extreme Learning Machine (DOS-ELM) with double dynamic forgetting factors (DDFF) and Updated Selection Strategy (USS) is proposed. First, depending on the characteristics of the training data that comes from aircraft sensors, robust feature extraction using a modified Denoising Autoencoder (DAE) is introduced to learn important patterns from data. Then, USS is integrated to ensure that only the useful data sequences pass through the training process. Finally, OS-ELM is used to fit the non-accumulative linear degradation function of the engine and to address dynamic programming by trucking the new coming data and forgetting gradually the old ones based on the proposed DDFF. The proposed DOS-ELM is tested on the public dataset of commercial modular aeropropulsion system simulation (C-MAPSS) of a turbofan engine and compared with OS-ELM trained with ordinary Autoencoder (AE), basic OS-ELM and previous works from the literature. Comparison results prove the effectiveness of the new integrated robust feature extraction scheme by showing more stability of the network responses even under random solutions. • Denoising OS-ELM with double dynamic forgetting factor and updated selection strategy. • Robust feature extraction using a modified denoising autoencoder. [ABSTRACT FROM AUTHOR]

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. [ABSTRACT FROM AUTHOR]