Accurate real-time fault detection, localization, and classification techniques are necessary to maintain grid stability and prevent faults. Traditional techniques have low accuracy rates, with high latency, and are generally incapable of handling the inherent non-linearity and dynamicity of power systems. Bearing these limitations in mind, we propose the novel Multi-Stage Deep Learning Framework, which significantly improves the process of fault management. First, it uses a hybrid Transformer-Convolutional Neural Network (CNN) in the process for fault detection. This is because transformers will capture the long-range temporal dependencies by voltage and current data, while CNNs will extract localization fault features. This approach can achieve an estimated fault detection accuracy of 99.3% with a real-time latency of less than 50 ms. For fault localization, a Graph Neural Network (GNN) is employed for modeling the power grid as a graph, hence enabling learning of spatial dependencies between nodes. This approach achieves 97.80% accuracy in the detection of exact fault location and completes within 100 ms. Next, the classification of faults uses an RNN-LSTM network to overcome temporal dependencies in the evolving nature of fault signals. The model reaches 98.20% classification accuracy by completing within 120 ms, which signifies that the model could differentiate between various kinds of faults in real-time applications. This framework improves the accuracy and speed of the fault management process, which directly contributes to the resilience and reliability of the power transmission system, thus reducing downtime period and increasing operation efficiency.

Validating Electronic Control Units (ECUs) for vehicle dynamics normally depends on real vehicles, which makes testing costly, time-consuming, and difficult to repeat. To address these issues, a simulation framework has been developed in the CANoe (Controller Area Network open environment) that combines CAPL (Communication Access Programming Language) scripting with Panel Designer. This setup reproduces key driving operations-such as gear shifting, braking, and creeping-and displays system responses through an interactive interface. Unlike earlier approaches that relied only on CANoe and CAPL, the inclusion of Panel Designer provides both input generation and clear output visualization within the same environment. A comparison with manual vehicle-level testing highlights improvements in flexibility, repeatability, and efficiency, while reducing dependence on physical prototypes. The framework offers a practical tool for engineers to accelerate ECU validation and strengthen safety assessments in the development of advanced and autonomous vehicles.

Feature representation techniques inherently introduce computational overhead, and conventional feature selection methodologies often discard closely correlated attributes, deeming them redundant. This study leverages Portable Executable Header (PEH) characteristics to construct an enriched feature representation, ensuring the preservation of critical and distinctive information while facilitating efficient extraction. A preliminary evaluation is conducted across six machine learning (ML) classifiers to identify the two most effective models for malware detection applications. To enhance feature representation, an advanced data preprocessing pipeline is employed prior to feature selection (FS). A Relief-F-based filtering mechanism is utilized to assign weighted importance to individual features, thereby preserving all relevant information. Iterative training with various weighted feature subsets enables the identification of an optimal, compact feature subset, denoted as A_ofs . The proposed methodology achieves a substantial reduction of feature overhead by 79.7

Magnetic Resonance Imaging (MRI) is a powerful diagnostic tool, but its slow acquisition speed poses challenges in clinical efficiency, cost, and patient comfort. Deep learning (DL) has emerged as a transformative approach to reconstruct high-quality MR images from undersampled k-space data, significantly accelerating scan times without sacrificing image fidelity. This paper explores recent advances in DL-based MRI reconstruction, emphasizing the need for robust, efficient, and clinically viable solutions. We examine key aspects such as architectural innovations, loss functions, data consistency enforcement, and hybrid-domain modeling, while highlighting the limitations of conventional evaluation metrics like SSIM and MSE. Emphasis is placed on diagnostic accuracy, robustness to noise and artifacts, and adaptability across different scanners and protocols. In the context of CyberPhysical Systems (CPS), we envision integrating these DL models with IoT-enabled healthcare platforms, enabling real-time, intelligent imaging workflows. We also address data scarcity and transfer learning as enablers for broad clinical adoption. By aligning deep learning strategies with CPS principles, this work contributes toward building scalable and sustainable medical imaging systems. Our findings aim to support the development of next-generation smart healthcare infrastructure by bridging advanced computation with clinical utility.

The severe threat posed by the pandemic of Severe Acute Respiratory Syndrome—Coronavirus (SARS-CoV2/COVID-19) to the health as well as economic living society across the world has led the countries been struggling with ways to stop the spread of the pandemic. The cases have continued to spread at an exponential rate across 218 different countries. Statistics indicate that number of people infected with the disease are 29.3 million in India alone with 175 million global cases. The death toll has also gone up to 363K in India and 3.78 million worldwide. In our study, techniques based on deep CNNs to detect people affected by COVID using real-world datasets are discussed. To identify patients infected with COVID-19 from healthy patients, we took chest X-ray image data. Many published papers indicate that they have acquired over 97

Nonlinear ultrasonic (NLU) has been established as an effective method for the non-destructive evaluation of power plant materials for various types of damage, including estimation of creep damage. However, the information obtained through NLU measurement may not be sufficient to predict either the failure probability or remaining creep life of any power plant component. A procedure has been formulated to estimate the probability of failure vis-a-vis creep life of power plant materials through a two-parameter Weibull analysis of NLU data. The investigation involved creep testing of P92 steel at 625 °C for three different applied stresses 120, 140, and 160 MPa. Subsequently, the extent of damage was estimated using Weibull distribution analysis from NLU parameter β , measured in the same specimen at different interruptions. The variation in cumulative distribution function (CDF) and the damage accumulation rate with increased damage, were examined. Further, the behavior of predicted NLU parameter β obtained using inverse CDF was evaluated with respect to measured β . Damage accumulation during creep deformation was confirmed through significant microstructural changes such as the growth and coarsening of precipitates, micro-crack formation, and their coalescence. Weibull distribution-based analysis established its potential as an alternative method for predicting the failure probability and life of power plant components under creep deformation from the NLU measurements.

Reliable and accurate prediction of the creep life of power plant components is crucial for both economic and safety reasons. Existing prediction models, based on creep test data, can be complex and time-consuming. Nonlinear ultrasonic (NLU) is a widely-accepted non-destructive testing (NDT) technique for evaluating damage progression in crept specimens. The information from NLU measurements alone is insufficient to forecast the life of any component. In real-time applications, intelligent NDT protocols are needed to enable fast and accurate life prediction of such components. A methodology for creep life prediction using artificial neural networks (ANN) has been introduced based on NLU test results of crept P92 steel specimens. The technique involved creep tests of P92 specimens exposed to a temperature of 625SUPERSCRIPT ZEROC with applied stress ranging from 120MPa to 160MPa, NLU measurements at each step load, and prediction of creep life of the material with a ANN trained with creep strain and NLU test data. The technique involves prediction from previously generated historical data, thus saving both cost and time of conducting continuous experiments. This approach for ANN modeling of NLU data can be considered a reliable, time-saving, and effective technique for assessing creep damage progression in power plant components.

Parkinson’s disease (PD) affects over a dozen million people worldwide. There is presently no solution for this, much to the dismay of these sufferers. This needs early detection for enhancing the patient’s overall well-being. The goal of our research is to create a prototype tool that can detect PD symptoms and assess their severity rate (dependent on the UPDRS) using gait signals. The projected model beats state-of-the-art approaches accomplishing an accuracy of 98.7

Several popular time-frequency techniques, including the Wigner-Ville distribution, smoothed pseudo-Wigner-Ville distribution, wavelet transform, synchrosqueezing transform, Hilbert-Huang transform, and Gabor-Wigner transform, are investigated to determine how well they can identify damage to structures. In this work, a synchroextracting transform（SET） based on the short-time Fourier transform is proposed for estimating post-earthquake structural damage. The performance of SET for artificially generated signals and actual earthquake signals is examined with existing methods. Amongst other tested techniques, SET improves frequency resolution to a great extent by lowering the influence of smearing along the time-frequency plane. Hence, interpretation and readability with the proposed method are improved, and small changes in the time-varying frequency characteristics of the damaged buildings are easily detected through the SET method.

Day by day, because of the increase in the population, energy use also increases, leading this generation towards the end of Non-Renewable sources. A free quotation which is completely unutilized, is available, but very few people are interested in using it, which is Solar Energy. In Solar Energy, this generation can have one-time investments and utilize a free energy source for a long time. The main problem is trust and past investments, which few governments support by giving multiple subsidies and benefits to those who want to use them. In this chapter, the author enlightens the technical know-how of Solar energy systems and their advantages and disadvantages from a technical perspective.

At present, ATMs (Automated Teller Machines) are one of the essential services for our daily life. It is also true that the thefts of false transactions and pin thefts are increasing yearly. A significant amount of theft at ATMs is due to pin overlooking and card skimming. Biometrics provide promising security but have high implementation costs. Also, Indian laws discourage using BiometricsBiometrics in all places. So, can AI be the solution to this problem? Instead of using keypad-based inputs for pins, gesture detection with AI can be used for secure inputs. A trained deep neural network can detect count from the hand symbols/gestures. The gesture input is given by inserting the hand inside a safe box with a high-resolution camera attached. The camera takes images and sends them to Raspberry Pi or any other embedded system. The Raspberry Pi executes the lightweight ML model to detect the count. The detected count is then encrypted and passed to the ATM. Using a gesture identification system removes the problem of pin theft and can be developed and implemented with the slightest modification in ATMs. In the current COVID period, execution of ATM works with minimum contact to public surfaces has increased immensely. In this system, a keypad is also removed and can further be incorporated to read a variety of inputs from gestures instead of just hands. This chapter explores how lightweight neural networks can be trained to detect sensors and run on low-processing systems like Raspberry Pi. We achieved an accuracy of 94%-97% in detecting gestures and pins where accuracy varies for each motion.

Industrial machine-vision (MV) applications require high-speed stitching of low-textural images from multiple high-resolution cameras for Field-of-View expansion. The most vital step in the stitching process is the effective and efficient extraction of features, which becomes challenging for low-textural images. This paper presents a comparative study of five popular feature descriptor algorithms for image stitching viz. Scale Invariant Feature Transform (SIFT), Speeded Up Robust Feature (SURF), Oriented Fast and Rotated BRIEF (ORB), Binary Robust invariant scalable keypoints (BRISK), and Accelerated-KAZE (AKAZE). The focus of this paper is to present a study of the performance comparison among these feature extraction methods for low-textural images from real-time steel surface inspection systems. Primarily, synchronized images of steel rolled at room temperatures are obtained from a two-camera network with overlapping regions. Feature descriptor algorithms extract features from two images with an overlapping area and further match the features using K-Nearest Neighbour (KNN) algorithm. The performance of the five feature descriptor algorithms is evaluated using a low-textural dataset that consists of a set of 177 images captured from two cameras placed at a fixed distance from each other. The efficiency of these algorithms is quantitatively and qualitatively evaluated using execution time, sensitivity, and specificity. Finally, this paper provides guidelines for future research on problems with FOV expansion in industrial scenarios.

Big Data is a massive collection of data that continues to grow dramatically over time. It is a data set that is so huge and complicated that no standard data management technologies can effectively store or process it. Big data is similar to regular data, but it is much larger. “There's no doubt that the volumes of data presently available are vast, but that's not the essential element of this new data ecosystem,” says one expert. The term “big data” is now commonly used to refer to the application of predictive analytics. New correlations can be discovered by analyzing data sets to “identify economic trends, prevent diseases, combat crime, and so on.” In sectors such as Internet searches, financial technology, healthcare analytics, geographic information systems, urban informatics, and business informatics, scientists, corporate executives, medical practitioners, advertising, and governments all face challenges with enormous data sets.

Day by day, because of the increase in the population, energy use also increases, leading this generation towards the end of Non-Renewable sources. A free quotation which is completely unutilized, is available, but very few people are interested in using it, which is Solar Energy. In Solar Energy, this generation can have one-time investments and utilize a free energy source for a long time. The main problem is trust and past investments, which few governments support by giving multiple subsidies and benefits to those who want to use them. In this chapter, the author enlightens the technical know-how of Solar energy systems and their advantages and disadvantages from a technical perspective.<br>

Electrical power distribution network is an integral part of electrical power systems since it is the last stage in the delivery of electricity to customers. The distribution network is responsible for distributing power to consumers at desired voltage levels with higher reliability. Alternating Current (AC) three phase four-wire structure is the standard distribution system that exists throughout the world. With the growth in urban population and development of industries, distribution grids now consider a considerable amount of power. The large number of lines in a distribution system experience regular faults which lead to high values of line currents. In the present work, an alternate solution to the problems related with interruptions by means of a statistical modeling of current sample database is applied to determine the fault location in power distribution systems in order to reduce the system restoration time. The current samples collected from the sample distribution systems are subjected to FCM to obtain clusters and fed to expectation–maximization algorithm. It gives an edge over the conventional impedance-based methods and also the problem of multi estimation has been successfully dealt.

To deal with the huge amount of data, minimizing the overhead will play a key role in speedy and efficient malware detection. We propose a machine learning (ML) malware detection model with preprocessing to limit the feature overhead. The portable-executable (PE) header information that retains meaningful and distinctive information has been considered to classify benign and malware files. The dataset is preprocessed by applying transformation, outlier detection and filling, and smoothing techniques. A maximum relevance minimum redundancy-based feature selection method is deployed to assign the rank and score to each feature retaining the maximum relevant and minimal redundant information. Based on the obtained rank, many subsets of features have been created and investigated against support vector machine (SVM) and k-nearest neighbors (k-NN) with parametric tuning. The proposed ML model integrated with data preprocessing, feature selection, and SVM-polynomial classifier has superior performance. This model is eliminating 63.8% feature overhead with accuracy above 99.1% for the benchmark datasets. To examine the robustness of the proposed model, new balanced and imbalanced datasets are created using new malware. The test results are encouraging with accuracy and specificity above 96.68%, 97.65%, and 91.57%, respectively. Interestingly, the proposed model is not trained using the newly created dataset.

The knowledge of remaining useful life and the probability of failure at any point of time in the life cycle of any power plant component is an important information for the plant operators to take preventive action. This paper focuses on life data evaluation of creep-exposed power plant material based on statistical probability distribution through Weibull analysis. The probability distribution was obtained considering the change in non-linear ultrasonic (NLU) parameter measured in P92 steel at different creep test conditions. The material was creep tested at 650 degrees C for three different applied stresses. The NLU parameter (beta), which indicates the extent of damage, is the ratio of the amplitude of the fundamental frequency of the transmitted signal to the square of the amplitude of the second harmonic of a sinusoidal wave propagated into the material. The two-parameter based Weibull distribution function was adopted for evaluating the cumulative distribution function and failure rate. A sudden increase in NLU parameter was observed at similar to 80% of creep damage followed by a drop in its value indicating the specimen failure. With increase in applied stress, failure rate increase was also observed. Microstructural observations revealed that with creep progress, the growth and coarsening of precipitates, micro crack formation and their coalescence were the major cause for increase in failure rate. Therefore, application of this technique can be useful for evaluating the creep life and probability of failure of any plant component in a non-invasive way. Copyright (C) 2022 Elsevier Ltd. All rights reserved. Selection and peer-review under responsibility of the scientific committee of the Third International Conference on Recent Advances in Materials and Manufacturing 2021.