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

 

Advanced Transformer Fault Prediction via LSTM and Digital Twin Integration|Chapter 8 | Engineering Research: Perspectives on Recent Advances Vol. 9

Transformers are the heart of electric power systems, and their operational state decides whether or not the power network is well-regulated. Electrical, mechanical, and thermal stresses cause some gases created during an operation to dissolve in insulating oil. The most significant tool for defect diagnostics in transformers is dissolved gas analysis (DGA). The time series prediction of dissolved gas levels in oil, when combined with dissolved gas analysis, provides a foundation for transformer fault diagnosis and an early warning. A long short-term memory (LSTM) based prediction model is developed in this paper to train the digital twin for identifying the essential fault in the transformer via DGA. The model is fed with three different gas concentrations as input. This study achieves the performance evaluation in terms of validation accuracy. The suggested model exhibits significant validation accuracy of 99.83%, as indicated by the analyses, thus aiding the early prediction of transformer maintenance. It can be validated that the LSTM model for fault identification and analysis using dissolved gas in the transformer has a lot of research potential. The study concluded that the trained digital twin integrated with the test transformer's condition monitoring system can precisely envisage the transformer's useful life. Its application in transformer online monitoring using a mobile device can be investigated.

 

Author(s) Details

GVSSN Srirama Sarma
Department of Electrical and Electronics Engineering, Matrusri Engineering College, Hyderabad, India.

 

Bumanapalli Ravindranath Reddy

Deputy Executive Engineer, Jawaharlal Nehru Technological University Hyderabad (JNTUH University), Hyderabad, India.

 

Pradeep Nirgude
Ultra High Voltage Research Laboratory (UHVRL), Central Power Research Institute (CPRI), Hyderabad, India.

 

 

Please see the book here:- https://doi.org/10.9734/bpi/erpra/v9/4142

Advances in AI and Machine Learning for Skin Disease Prediction | Chapter 16 | Innovative Solutions: A Systematic Approach towards Sustainable Future

The field of dermatological diagnosis has made significant strides recently with the application of AI and machine learning approaches. The aim of the study is to investigate the potential of AI and machine learning technologies in diagnosing and predicting skin diseases. Research has utilized techniques including convolutional neural networks, deep learning, and ensemble data mining to identify a variety of skin disorders with high accuracy rates, frequently between 88 and 100 Percentage. Even with these advancements, there are still issues to be resolved, such as the scarcity of datasets and uncertainties regarding practicality. Still, there is great potential for revolutionizing dermatological diagnoses with new methods that combine deep learning and computer vision. In order to really use these technologies in clinical settings, more testing and improvement are required.

 

Author (s) Details

 

Tharigopula Madhan Mohan
Department of Computer Science, Christ (Deemed to be a University), Bengaluru, India.

 

Smera C
Department of Computer Science, Christ (Deemed to be a University), Bengaluru, India.

 

Sandeep J
Department of Computer Science, Christ (Deemed to be a University), Bengaluru, India.

 

Sreeja CS
Department of Computer Science, Christ (Deemed to be a University), Bengaluru, India.

 

Karthik Raja
Department of Computer Science, Christ (Deemed to be a University), Bengaluru, India.

 

John Gilbert
Department of Computer Science, Christ (Deemed to be a University), Bengaluru, India.

 

Please see the book here:- https://doi.org/10.9734/bpi/mono/978-93-49238-47-3/CH16 

Electroencephalography-Based Inner Speech Classification Using LSTM and Wavelet Scattering Transformation (WST) | Chapter 3 | Contemporary Perspective on Science, Technology and Research Vol. 3

In this place paper, we propose an imagined talk-based idea pattern recognition using deep education. Multiple features were culled concurrently from eight-channel Electroencephalography (EEG) signals. Imagined talk, sometimes named inner speech, is an wonderful choice for decoding human thinking using the Mind-Computer Interface (BCI) idea. BCI is being developed to happening slowly allow paralyzed victims to interact directly accompanying their environment. To acquire classifiable EEG data with lean number of sensors, we placed the EEG sensors on carefully picked spots on the scalp. To decrease the dimensions and complicatedness of the EEG dataset and to avoid overfitting all along the deep learning invention, we utilized the wavelet scattering renewal. The study was conducted in the Department of Energetic & Computer Metallurgy and Computer Science at Jackson State University, United states of america. A low-cost 8-channel EEG headset was secondhand with MATLAB 2023a to get the EEG data. The Long-Short Term Thought Recurrent Neural Network (LSTM-RNN) was used to decipher the identified EEG signals into four audio commands: Up, Below, Left, and Right. Wavelet uneven transformation was applied to extract ultimate stable features avoid the EEG dataset through a series of filtration processes. Filtration has been executed for each individual command in the EEG datasets. The projected imagined speech-located brain wave pattern recognition approach obtained a 92.50% overall classification veracity. This accuracy is promising for plotting a trustworthy imagined talk-based Brain-Calculating Interface (BCI) future authentic-time systems. For better judgment of the classification performance, additional metrics were considered, and we got 92.74%, 92.50% and 92.62% for precision, recall, and F1-score, individually. Future work is planned to implement and test an online BCI order using MATLAB/Simulink and G. tec Unicorn Mixture Black+ bias.

Author(s) Details:

Mokhles M. Abdulghani,
Jackson State University, MS, USA.

Wilbur L. Walters,
Jackson State University, MS, USA.

Khalid H. Abed,
Jackson State University, MS, USA.

Please see the link here: https://stm.bookpi.org/CPSTR-V3/article/view/12915 

Prediction of Geological Parameters during Tunneling by Time Series Analysis on In-situ Data: A Recent Study | Chapter 06 | Novel Perspectives of Engineering Research Vol. 10

 A tunnel boring machine (TBM) is a piece of heavy-duty equipment used to construct underground tunnels. Geological conditions are crucial components in the tunnelling process, as they have a direct impact on the control of construction equipment. The goal of this work is to use continually gathered aerial characteristics to anticipate geological information ahead of TBM. This study focused on the underlying links between the sequential nature of tunnel in-situ data and the continuous interaction between equipment and geology, and used the long short-term memory (LSTM) time series neural network technique for processing in-situ data. A method for predicting geological characteristics in advance is proposed based on TBM real-time status monitoring data. The proposed method was used to predict five geological parameters for a tunnel project in China, with R2 values more than 0.98 for all five geological parameters. In terms of performance, the LSTM was compared to an artificial neural network (ANN). The LSTM's prediction accuracy was significantly higher than the ANN's, and its generalisation and robustness were also better than the ANN's, showing that the recommended LSTM approach could extract the in-situ data's sequence features. The rule of equipment-geology interaction was mirrored in the model's memory structure thanks to the adoption of the "gate" principle, allowing for precise prediction of geological parameters during tunnelling. The model's impact on the time frame and prediction distance is also investigated. The proposed method provides a novel approach to obtaining geological data during TBM construction, as well as a framework for analysing in-situ data with sequence properties.

Author(S) Details

Shanglin Liu
Key Laboratory of Modern Engineering Mechanics, Tianjin University, Tianjin 300072, China.

Kaihong Yang
Key Laboratory of Modern Engineering Mechanics, Tianjin University, Tianjin 300072, China.

Jie Cai
Design and Research Institute of Tunneling Machine, China Railway Construction Heavy Industry, Changsha 410100, China.

Siyang Zhou
Key Laboratory of Modern Engineering Mechanics, Tianjin University, Tianjin 300072, China.

Qian Zhang
Design and Research Institute of Tunneling Machine, China Railway Construction Heavy Industry, Changsha 410100, China.

View Book:- https://stm.bookpi.org/NPER-V10/article/view/6507