A Customised LSTM-Based Deep Learning Framework for Transformer Predictive Maintenance: Performance Analysis | Chapter 5 | New Horizons of Science, Technology and Culture Vol. 9

 

Transformers are critical and costly components of power systems whose health deteriorates over time due to factors such as poor cooling and heavy loading. Consequently, predictive maintenance is emerging as an effective alternative to conventional corrective maintenance, enabling continuous monitoring and early fault detection.

 

To enhance the effectiveness of predictive maintenance for power transformers under limited Dissolved Gas Analysis (DGA) data conditions, this study proposes a customised Long Short-Term Memory (C-LSTM) deep learning model. The developed C-LSTM architecture is specifically designed to address the limitations of conventional LSTM networks, which often exhibit higher classification error rates when trained on small datasets and may underperform compared to traditional machine learning approaches.

 

A comprehensive performance evaluation was conducted by comparing the proposed C-LSTM model with several well-established traditional machine learning algorithms using multiple metrics, including validation accuracy, test accuracy, precision, recall, and F1-score. Additionally, the diagnostic capability of the model was rigorously assessed across seven transformer fault categories, including low- and high-energy discharges, partial discharge, electrical and thermal faults, and low-, medium-, and high-temperature thermal faults.

 

The experimental results demonstrate the superior classification and diagnostic performance of the proposed C-LSTM model, achieving a validation accuracy of 100% and a test accuracy of 98.57%, significantly outperforming conventional machine learning techniques. These findings confirm that the proposed C-LSTM framework offers a robust and reliable solution for transformer fault diagnosis and predictive maintenance, particularly in scenarios characterised by scarce DGA datasets.

 

 

Author(s) Details

G.V.S.S.N. Srirama Sarma
Department of Electrical and Electronics Engineering, Matrusri Engineering College, Saidabad, Hyderabad, India.

 

Please see the book here :- https://doi.org/10.9734/bpi/nhstc/v9/6804

 

Sequential Mathematical Programming With \(\zeta\)-Analysis for HESVM | Chapter 9 | Mathematics and Computer Science: Contemporary Developments Vol. 7

Vapnik's quadratic programming (QP)-based support vector machine (SVM) is a state-of-the-art powerful classifier with high accuracy while being sparse. Moving one step further in the direction of sparsity, Vapnik proposed one more SVM that uses linear programming (LP) for the cost function. This machine, compared with the complex QP based one, is more sparse but offers similar accuracy, which is essential to work on any large dataset. However, further sparsity is optimum for computational savings as well as to work with very large and complicated datasets. Producing even more sparsity without reducing generalization capability of a detector is extremely challenging. In this dimension, we apply a distinct sequence of Mathematical Programming followed by slack variable analysis that leads to an exceptionally fast and accurate SVM based detector. Being immensely sparse and optimally complex, this Highly Effcient SVM (HESVM) can expertly work on very large and noise-effected complicated data. Experiments on Benchmark data shows that HESVM requires kernel execution as little as 6.8% of the classical QP based SVM while producing nearly the same classification accuracy on test data and demanding 42.7, 27.7 and 46.6% that of other three executed cutting-edge heavy-sparse machines while posing similar classification accuracy. It also claims the least Machine Accuracy Cost (MAC) value among all of these machines though producing very similar generalization performance, which is calculated statistically using the term Generalization Failure Rate (GFR). Being quite practical for contemporary technological development, it has become indispensable for optimum manipulation of the troublesome massive, and diffcult data.

 

Author (s) Details

 

Rezaul Karim
Uttara University, Bangladesh.

 

Amit Kumar Kundu
Uttara University, Bangladesh.

 

Ali Ahmed Ave
Uttara University, Bangladesh.

 

Please see the book here:- https://doi.org/10.9734/bpi/mcscd/v7/2725

Application of Machine Learning for the Tracing of Jellyfish Attack in MANETs | Chapter 4 | Science and Technology: Developments and Applications Vol. 2

Mobile ad-hoc networks (MANETS) are one of the emerging fields that have seamless applications in the field of emergency situations like rescue operations, commercial applications like virtual classrooms, medical like disease diagnosis etc. In the era of driverless vehicles, mobile ad-hoc network (MANET) finds a useful place in discoveries. However, MANETS are vulnerable to a number of attacks because of properties like non-existing infrastructure, dynamic topology, multihop network etc. A lot of previous works have focused on the impact of various attacks on routing protocols like jellyfish attacks, blackhole attacks or selfish node attacks. The area becomes more vulnerable to attacks as a network is in use for a limited short time as the topology is highly dynamic and time-specific. However, the use of machine learning and deep learning algorithms has given a new edge to MANETS in driverless vehicles. In literature, malicious node/ selfish node detection (passage of wrong information/ blockage of information) was done by using various supervised or unsupervised machine learning algorithms like KNN, SVM, CNN or recurrent networks. This paper presents the detection of various faulty nodes using the NS2 simulator and using machine learning algorithms. The study concluded that the application of machine learning and deep learning algorithms has increased the faulty node detection accuracy as compared to simulated code. The increased accuracy may be helpful in the formation of short-term and highly dynamic MANETS like driverless cars or military networks. In future, these networks may be connected with satellite images so as to forecast whether information like various DANA in Orissa, and SARA in US pandemics and take precautionary actions accordingly.

 

Author (s) Details

Bhawna Singla
Geeta University, India.

A. K. Verma
Thapar University, India.

 

Please see the book here:- https://doi.org/10.9734/bpi/stda/v2/3679

Detection of Deceit Using Speech and Thermal Videos During Interactive Sessions: An Integrated Analysis | Chapter 10 | Advances and Challenges in Science and Technology Vol. 4

 This study reduced on the use of interrogation visual and audio entertainment transmitted via radio waves and thermal imaging to label deception. In the study, a real-life table has been generated by constituting a real corruption of stealing scenario. During the questioning process the thermal broadcast and audio records have been carried out together in a concealed tone. The police area has long used polygraphy as the gold standard for storyteller detection. In due course, various contemporary methods for lying or deception discovery have emerged that are more correct and straightforward. One of ruling class is the analysis of thermal television. Thermal footage is written during an interrogation and following examined to expect signs of lying. Additionally, it is attainable to assess a person's performance from the audio record, which maybe a crucial hint for uncovering deceit. The temperature of the brow and periorbital areas is culled to measure the blood flow rate. It was observed that either of person the one is lying, the temperature of these regions increased in addition to those of truth-tellers. Analysis of the talk of the subjects was also completed activity to understand the distinctness in the pattern of speech attributes of storytellers and truth-tellers. It was found that storytellers take more silent pauses and have taller values of the help out their voice when they were being interrogated. Finally, we have combined the effect of thermal television and audio study for decision making.

Author(s) Details:

Saswata Satpathi,
Department of Electrical Engineering, Indian Institute of Technology, Kharagpur, West Bengal, India.

Pooja Kumawat,
Department of Electrical Engineering, Indian Institute of Technology, Kharagpur, West Bengal, India.

Aurobinda Routray,
Department of Electrical Engineering, Indian Institute of Technology, Kharagpur, West Bengal, India.

Partha Sarathi Satpathi,
Department of Microbiology, Midnapore Medical College and Hospital, Midnapore, West Bengal, India.

Please see the link here: https://stm.bookpi.org/ACST-V4/article/view/12102

Image Processing Techniques for Medical Mammography Masses Detection Applications | Chapter 12 | Research Highlights in Science and Technology Vol. 9

The function of image refine is crucial in the mammography for detecting of any undesired suspicious domains in the mammogram. Quality of image is majorly met in proposed concept enhancement technique for bringing back of efficient feature extractions. SVM and GMM located classifier models are built for reconstructing the performance limits of proposed methodologies. Aim concerning this chapter search out reduce the commotion in the mammogram image in order to embellish the mammogram image. Enhancement methods, such as, contrast elongated and histogram counterweight, morphological, median filters are investigated on mammogram images and results are distinguished for analyzing the representation enhancement. These techniques are used to advance the clarity and denoise of mammogram concepts. It is useful for detecting the tumors in the mammogram efficiently and it is helpful to establish and size of masses or tumors correctly in mammogram. Proposed methodology shows the figure from the selected range of pixels and it show the force of the mammogram image at the indicated pixel range. By this, the mammogram image will be smoothened that will be useful for diagnosing the tumors.

Author(s) Details:

K. Rajendra Prasad,
Department of CSE (CS), Institute of Aeronautical Engineering, Hyderabad, Dundigal, Telangana-500 043, India.

Please see the link here: https://stm.bookpi.org/RHST-V9/article/view/11680

An Efficient Machine Learning Model for Prediction of Dyslexia from Eye Fixation Events | Chapter 15 | New Approaches in Engineering Research Vol. 10

 Dyslexia is not curable, although dyslexics can achieve great success in school and in life with the right remedial support. Eye movement patterns during reading can help you understand dyslexia-related reading issues better. An eye-tracker can be used to capture eye movements and deduce the relationship between how eyes move in response to the words they read. In this study, raw eye tracking data was used to construct a collection of binocular fixation and saccade features based on statistical measurements. In this study, raw eye tracking data was used to construct a collection of binocular fixation and saccade features based on statistical measurements. To develop classification models for dyslexia prediction, machine learning techniques such as the Random Forest Classifier (RF), the Support Vector Machine (SVM) for classification, and the K-Nearest Neighbor (KNN) for prediction of dyslexia were examined. KNN exhibited 95 percent accuracy over a limited feature set related with fixations and saccades, compared to SVM and RF. These properties of the eyes can be used to create dyslexia prediction screening tools. Early discovery of dyslexia can help children get the treatment they need, helping them to succeed in school.

Author (S) Details

A. Jothi Prabha
Jyothishmathi Institute of Technology & Science, India.

R. Bhargavi
Vellore Institute of Technology University, Chennai, India.

B. Harish
VIT University, Chennai Campus, India.

View Book :- https://stm.bookpi.org/NAER-V10/article/view/2991

Study on Opinion Mining Framework Using Proposed RB-Bayes Model for Text Classification| Chapter 8 | New Approaches in Engineering Research Vol. 9

 Everyone is on social media, and their motivation is not only to be active but also to generate knowledge. We always read reviews on social media before making a purchase. Information mining is a capable concept with enormous potential for predicting future patterns and behaviour. It refers to the process of extracting hidden information from large data sets using techniques such as factual investigation, machine learning, grouping, neural systems, and genetics.algorithms. There is a problem of zero likelihood in naïve bayes. To solve the problem of zero likelihood, this work proposed the RB-Bayes technique, which is based on Baye's theorem. We also compare our strategy to a few other approaches, such as naive bayes and SVM. We show that this technique is superior to several existing strategies, and that it can analyse data sets more effectively. When the proposed approach is applied to real-world data sets, the results are generally more accurate. The precision of the RB-Bayes computation is 83,333.

Author (S) Details

Dr.Rajni Bhalla
Department of Computer Application, Lovely Professional University, India.

Dr. Amandeep Bagga
Department of Computer Application, Lovely Professional University, India.


View Book :- https://stm.bookpi.org/NAER-V9/article/view/2821

Obstacles Detection at a Railroad Crossing Using the Histogram of Oriented Gradients Method and Support Vector Machine Classifier | Chapter 5 | New Approaches in Engineering Research Vol. 8

 The goal of this project was to use the histogram of oriented gradients approach and a support vector machine to detect impediments at railroad crossings. A railroad crossing is a location where train tracks cross other routes, such as a highway. Railroad crossings must be equipped with signs, markers, traffic signalling systems, and crossing gate guards, according to Minister of Transportation Regulation Number 36 of 2011. 3477 of the 4716 level crossing points, on the other hand, lack a railroad keeper, making them vulnerable to traffic accidents. Furthermore, at night and in cloudy conditions, hazard information (warning signals) from the railroad keeper to the OOperation Center and machinists can be difficult to view. As a result, the goal of this study is to use the Histogram of Oriented Gradient (HOG) method and the Support Vector Machine (SVM) classifier to detect obstacles (cars) at a railroad crossing. HOG is in charge of extracting object characteristics (cars), whereas SVM is in charge of classifying car objects based on whether or not they meet the criterion for car features. The results show that automobile objects had an accuracy rate of 85 percent, unoccupied train tracks had a rate of 73 percent, and passing trains had a rate of 91 percent.

Author (s) Details

A. Sugiana
School of Electrical Engineering, Telkom University, Bandung, Indonesia.

B. S. Aprillia
School of Electrical Engineering, Telkom University, Bandung, Indonesia.

M. N. Rifqi
School of Electrical Engineering, Telkom University, Bandung, Indonesia.

View Book :- https://stm.bookpi.org/NAER-V8/article/view/2270

Application of EEG Signals – A Case Study | Chapter 10 | Advanced Aspects of Engineering Research Vol. 11

 Traditional iterative estimate methods are replaced by the Gauss-Jacobi combinatorial algorithm. In nonlinear models, where other parameter estimate approaches fail, the combinatorial approach is frequently employed for outlier diagnosis. The purpose of this study is to compare the effectiveness of the Gauss-Jacobi and Gauss-Markov models when used to the parameter estimation process of a levelling network for the purpose of determining the efficiency of a combinatorial algorithm in a simple linear model. The Man-Machine-Interface (MMI) is a communication device that connects the brain to a computer in order to obtain and analyse brain data. The electrical signals produced by nerve cells are captured by the EEG. The goal of this work is to present the findings of EEG signal categorization and the use of appropriate music to represent different people's emotions based on their emotions. The study was carried out on a dataset that included 10 people who were exposed to patriotic, joyful, romantic, and sad songs, as well as relaxing activities.

Author (s) Details

Guruprasad S.
BMS Institute of Technology & Management, Bengaluru, Karnataka, India.

Veena N.
BMS Institute of Technology & Management, Bengaluru, Karnataka, India.

S. Mahalakshmi
BMS Institute of Technology & Management, Bengaluru, Karnataka, India.

View Book : https://stm.bookpi.org/AAER-V11/article/view/1255