Daily Archives: April 14, 2025

Smart Railway Safety and Track Monitoring System Using ESP32
Authors:- Assistant Professor Ambika Annavarapu, Vanja Chaitanya Aswith, Tadiboina Baji, Sattu Naga Likhith, Purushothapatanam Bhanu Sai Ram

Abstract- Railway transportation is one of the most widely used and efficient means of transport worldwide. However, railway accidents due to track failures pose a significant threat to human lives and infrastructure. Cracks and irregularities in railway tracks can cause derailments, leading to catastrophic accidents. To address this problem, we propose a Smart Railway Safety and Track Monitoring System that detects track anomalies using ultrasonic sensors. And ultrasonic sensors are connected to 180° servo motor. The system is designed to continuously monitor the track condition, and if the distance between the track and the ultrasonic sensor exceeds 5 cm, it Indicates a crack or damage. Upon detection of an anomaly, the system triggers a series of safety measures: it sends an alert to the nearest railway station via the Blynk IoT app, transmits a text message with GPS coordinates using a GSM module, and initiates an automated phone call to the concerned authority. Additionally, signal lights positioned beside the track serve as a visual warning, turning from green to red upon detection of a crack. The entire process is controlled by an ESP32 microcontroller, ensuring real-time monitoring and efficient safety measures.

DOI: 10.61137/ijsret.vol.11.issue2.334

Published by: Kajal Tripathi

Engineering Scalable Microservices: A Comparative Study of Serverless Vs. Kubernetes-Based Architectures
Authors:- Sreenivasulu Navulipuri

Abstract- This study compares serverless architectures and Kubernetes-based orchestration systems in the context of cloud-native microservices. It examines key factors such as latency, scalability, cost efficiency, and operational complexity. AWS Lambda and other serverless platforms scale automatically but experience delays during cold starts while Kubernetes provides detailed control at the cost of increased operational expenses. The paper also examines AI-driven resource optimization and hybrid models, such as Knative and OpenFaaS, which combine the advantages of both paradigms. Performance benchmarks and case studies guide architects in selecting the most suitable deployment model. A hybrid solution appears to provide the optimal combination of scalability, cost management and operational efficiency.

DOI: 10.61137/ijsret.vol.11.issue2.333

Published by: Kajal Tripathi

Evaluating Machine Learning Efficiency: Simpler Models Outperform Deep Learning in Motor Fault Detection
Authors:-Mrs. L. Yamuna., G. Abhisekhar., K. Prasanna Lahari., K. Vivek., S. Hemanth.

Abstract- In motor condition monitoring, deep learning techniques have been explored by utilizing two-dimensional plots as datasets instead of traditional time-series signals. For instance, Convolutional Neural Networks (CNNs) have been trained using recurrence and frequency-occurrence plots. While previous studies have shown promising results with CNNs, the indistinct differences in these plots often make the model’s decision-making process appear as a black box. This study evaluates and compares ten traditional machine learning (ML) techniques with recent deep learning (DL) approaches for motor fault diagnosis using the same dataset. The dataset consists of 3,750 synthetically generated motor current signal samples, categorized into five classes—one representing healthy conditions and four representing faulty motor conditions—each tested under five loading levels (0%, 25%, 50%, 75%, and 100%). Following similar training and testing phases, the Light Gradient Boosting Machine (LightGBM) achieved the highest classification accuracy of 93.20%, outperforming three CNN-based models by at least 10.4%, whose accuracy ranged between 74.80% and 82.80%. LightGBM also demonstrated superior performance in other key evaluation metrics, including F1 score, precision, and recall. Notably, five out of ten traditional ML models surpassed the CNN-based models. These findings emphasize the importance of carefully selecting deep learning architectures, as they are computationally expensive and memory-intensive, yet do not always guarantee improved performance over traditional ML models, especially for relatively straightforward tasks like motor fault classification using current signals.

DOI: 10.61137/ijsret.vol.11.issue2.328

Published by: Kajal Tripathi

AI-Driven Time Series Forecasting for Financial Markets: Leveraging Machine Learning for Smarter Predictions
Authors:-S.Likhita, S.Venkata Basavayya, Y.S.Santosh Kumar, P.Bhanu Divyasri, V.Sandeep, Mrs.V.Anantha Lakshmi

Abstract- Financial markets, including stock prices, exchange rates, and commodity prices, are inherently volatile and influenced by numerous factors, making their prediction a challenging yet essential task. Accurate forecasting of market trends is crucial for investors, financial analysts, and policymakers, as it helps in making informed decisions and mitigating risks. In this study, we explore the use of Support Vector Machine (SVM), a powerful machine learning algorithm, for time series forecasting of financial market trends. Traditional forecasting methods often struggle with financial data due to its non-linear and dynamic nature. However, SVM is well-known for its ability to handle high-dimensional data and capture complex patterns, making it a suitable choice for financial market prediction. Our approach leverages historical price and volume data to train the SVM model, enabling it to recognize patterns and predict future market movements. The study evaluates how effectively SVM adapts to changing market conditions, demonstrating its ability to model non-linear relationships within financial time series. Additionally, we consider external economic factors that may influence market behavior, further validating the robustness of the model. The findings highlight the potential of SVM in financial forecasting, offering a reliable alternative to traditional methods. Future work may involve integrating hybrid models combining SVM with deep learning techniques or incorporating macro-economic indicators to further enhance prediction accuracy. This research contributes to the growing field of AI-driven financial analysis, paving the way for more sophisticated and data-driven investment strategies.

DOI: 10.61137/ijsret.vol.11.issue2.329

Published by: Kajal Tripathi

AI-Powered Fraud Detection: Secure Online Transaction Monitoring Using Machine Learning
Authors:-G.Jashwitha, T.Sai Srinath, G.Naga Kastusi, V.Anshitha, G.Janitha Sree, Mrs.G.Tejasri Devi

Abstract-Fraud detection remains one of the most critical challenges in financial transactions, driving on going research and the adoption of advanced technologies such as machine learning. Financial transaction fraud detection aims to explore and compare various machine learning approaches to assess their effectiveness, challenges, and potential future developments comprehensively. This paper begins by highlighting the importance of fraud detection in financial transactions, emphasizing the widespread impact of fraudulent activities on individuals, businesses, and the overall economy. While traditional fraud detection methods have been valuable, they often struggle to counter increasingly sophisticated and evolving fraudulent schemes. As a result, more advanced techniques are required to enhance detection accuracy. Machine learning-based approaches have emerged as a promising solution, enabling algorithms to analyse vast amounts of transactional data and identify patterns indicative of potential fraud. In particular, supervised learning techniques—such as logistic regression, decision trees, and support vector machines—have gained significant popularity in fraud detection due to their ability to classify transactions as legitimate or fraudulent based on historical data.

DOI: 10.61137/ijsret.vol.11.issue2.326

Published by: Kajal Tripathi

Application of First Order Linear Ordinary Differential Equations in Mechanics and Thermodynamics
Authors:- Jyotika Sa, Tejaswini Pradhan

Abstract- This comprehensive study explores the profound applications of first-order linear ordinary differential equations (ODEs) in the domains of classical mechanics and thermodynamics. These mathematical tools serve as vital instruments in modeling and analyzing real-world physical phenomena. In particular, this research focuses on Newton’s Second Law of Motion and Newton’s Law of Cooling, both of which are quintessential examples of how first-order linear ODEs can effectively describe dynamic systems. The paper provides an in-depth explanation of the formulation, derivation, and solution of these equations, supported by descriptive illustrations and analytical interpretations. Emphasis is placed on demonstrating the solution techniques such as the integrating factor method, and the separation of variables method, while linking their mathematical elegance to practical engineering, environmental, and forensic applications. The ultimate objective is to illuminate how first-order linear ODEs not only simplify complex physical laws but also enable predictions that are essential in technological and scientific advancements.

DOI: 10.61137/ijsret.vol.11.issue2.332

Published by: Kajal Tripathi