Authors: J. Jenitta, Amulya N, Keerthi Kandakur, Nithin N, Shashank G

Abstract: As the online socialization process advances, more and more individuals are reporting their emotional conditions online and in many instances showing symptoms of misery and suicidal tendencies, which present significant obstacles to monitoring mental health. Conventional techniques of detecting such content are often untimely and ineffective and it is based on this need that automated and real time detection is important. In this paper, the author suggests a deep learning-based suicide risk prediction system based on linguistic data obtained on social media. The model achieves a high degree of accuracy in differentiating between texts that are suicidal and those that are not by learning the contextual semantics using bidirectional long short-term memory (Bi-LSTM) networks. The system is coupled with an IoT-based alert system to make sure that timely intervention is applied to prevent suicidal intent by sending alarms to caregivers or other interested parties whenever any suicidal intent is identified. The results of the testing show that the proposed method performs better in comparison to the traditional machine learning classifiers that are measured in terms of precision, recall, and F1-score. This combination related to the higher-order NLP techniques and the use of IoT-driven indicators offers a scalable and productive approach to stopping destructive online behavior. This article highlights the role of artificial intelligence and IoT in the improvement of digital mental health support systems.

DOI: http://doi.org/10.5281/zenodo.17590356

Published by: vikaspatanker

Authors: Hachimenum Nyebuchi Amadi, Henry Okechukwu Williams, Richeal Chinaeche Ijeoma

Abstract: The rapid integration of electric vehicle (EV) fast charging stations in DC micro-grids has introduced significant power quality challenges, particularly harmonic current distortion at the point of common coupling (PCC). In this study, a DC microgrid integrating photovoltaic (PV) generation, battery energy storage systems (BESS), and a Level-3 EV fast charging station was modeled in MATLAB/Simulink to examine the effect of harmonic distortion and evaluate mitigation using an Artificial Neural Network (ANN)-controlled Static Synchronous Compensator (STATCOM). Base case simulation results revealed that the EV fast charging station injected excessive harmonic distortion into the network, with dominant odd harmonics at the 11th and 13th orders, leading to a total harmonic distortion (THD) of 14.05%. This value significantly exceeds the IEEE 519-2022 standard limit of 8% for medium-voltage systems. Following the installation of an ANN-tuned STATCOM at the PCC, the harmonic distortion was substantially mitigated, reducing the 11th and 13th orders to 0.01% and 0.15% respectively. Consequently, the total harmonic distortion was minimized to 1.23%, achieving a 91.24% reduction and ensuring full compliance with IEEE standards. Furthermore, the ANN controller demonstrated excellent training performance with a best validation mean square error of 0.0034611 at epoch 20 and a regression correlation coefficient of R = 0.9879, validating its accuracy and robustness. These findings confirm that ANN-controlled STATCOM provides an effective and intelligent solution for enhancing power quality and system stability in DC micro-grids with EV fast charging integration.

DOI: http://doi.org/

Published by: vikaspatanker

Published by: vikaspatanker

Authors: Hachimenum Nyebuchi Amadi, Iyowuna Winston Gobo, Ugochi Benedicta Uche-Ibe, Richeal Chinaeche Ijeoma

Abstract: The reliance on fossil fuels and the need for effective battery management are significant challenges that renewable micro-grids seek to address. Fluctuations in supply and demand often result in higher operational costs and increased dependence on the external grid. With the urgent need to confront energy and environmental issues like global warming, transitioning to clean energy sources is becoming more viable. This study focuses on the Jetty 11kV feeder from the Abuloma 33kV injection substation in Port Harcourt, with an installed capacity of 1 x 7.5MVA. Currently, the feeder has a peak load of 3.9MW and an average load of 2.2 MW. To leverage the local abundance of water, the research aims to design a micro-grid using solar and wind energy. Using MATLAB Simulink, data from NASA meteorological sources will be simulated. The design features a 4.5MW photovoltaic (PV) array, a 2.5 MW wind energy source, and a 4 MWh battery storage unit. Despite variations in irradiance, significant improvements in power extraction were observed, with up to 4.5 MW generated by the PV array and 2.5MW by the wind turbine during peak times. The battery can be fully charged in four hours, and it was maintained at 40% capacity during low energy output periods. The State of Charge (SoC) of the battery showed dynamic behavior, enabling it to respond effectively to system imbalances and enhance microgrid resilience. A fuzzy logic controller (FLC) was used to manage charge and discharge cycles according to real-time parameters, ensuring reliable micro-grid operation even with low battery levels. The economic analysis revealed an initial cost of ₦990,251,352.19, a replacement cost of ₦414,669,375.12, a net present cost (NPC) of ₦10,813,540,000.00, and a levelized cost of electricity (COE) of ₦230.90/kWh. The low operation and maintenance (O&M) costs associated with renewable energy reduce reliance on the conventional grid and prolong infrastructure lifespan. Environmental assessments indicated a total greenhouse gas emission of 15,330,621 kg/year, significantly lower than that of conventional systems. The results confirm that the optimized hybrid renewable energy micro-grid enhances energy balance and resilience, showcasing its feasibility as a cost-effective and environmentally sustainable alternative to traditional power generation. The research aims to improve system resilience, reduce operating costs, and enhance micro-grid efficiency.

DOI: http://doi.org/10.5281/zenodo.17587580

Published by: vikaspatanker

Authors: Hachimenum Nyebuchi Amadi, Biobele A. Wokoma, Victor Nneji Chikwendu, Richeal Chinaeche Ijeoma

Abstract: A micro-grid is a localized energy system that typically operates as part of a larger, wide-area synchronous grid but can function independently when necessary. It comprises energy generators, loads, storage units, and control systems, all highly integrated and manageable. This study presents the design and implementation of a 60,000-watt solar photovoltaic (PV) microgrid incorporating an advanced fault detection and localization mechanism, aimed at addressing the limitations of conventional reactive fault systems. These traditional systems often respond only after fault currents surpass the tolerance thresholds of grid components, leading to reduced efficiency, equipment damage, or total system failure. To mitigate these issues, a DC micro-grid consisting of six solar PV arrays was modeled using Proteus 8.15 Professional and Siemens TIA Portal. Each array comprised 32 units of 400W, 12V panels arranged in an 8×4 configuration, delivering 72V per array. The PV arrays were individually connected through dedicated contactors (MCB1–6). Fault detection and isolation were achieved using smart electronics, specifically Arduino Nano microcontrollers integrated with WCS1600 current sensors capable of sensing up to 500A. The system efficiently identified and isolated faults occurring within any array. During testing, no faults were flagged for current values of 72.33A, 90.42A, 123.69A, 117.15A, 172.02A, and 199.09A, as they remained within the safe 200A threshold. However, overcurrent values recorded at PV arrays 3, 4, 5, and 6 (235.09A, 307.43A, 412.72A, and 209.09A, respectively) due to simulation of fault (short circuit, load-related faults, battery system faults, DC bus fault or converter and distribution fault) were promptly detected, and the affected arrays were disconnected to protect the system. Compared to previous research, this approach leveraging a hybrid of Arduino Microcontroller and Siemens S7-1200 PLC (CPU1214CDC/DC/DC) demonstrated improved efficiency and reliability in proactive fault detection and localization. Ultimately, the study successfully developed a programmable, feedback-enabled microgrid system capable of anticipating and mitigating faults before component tolerance limits are breached.

DOI: https://doi.org/10.5281/zenodo.17587332

 

Published by: vikaspatanker

Authors: M. Kumaraguru, B. Bhuvaneswari

Abstract: Cloud gaming is a rapidly evolving domain that provides seamless access to immersive, high-quality gaming experiences. Despite its advantages, reducing latency remains a significant hurdle, especially under varying network conditions. This study introduces an innovative solution that leverages artificial intelligence (AI) to tackle these issues. The proposed system integrates AI techniques to enhance multiple facets of cloud gaming, such as video compression, traffic routing, resource distribution, and prediction of user interactions. Machine learning algorithms continuously fine-tune streaming configurations in response to live network metrics and individual user preferences, thereby lowering latency and boosting visual fidelity. Furthermore, reinforcement learning is employed to optimize backend resource management, improving both scalability and operational efficiency. The use of AI-powered predictive analytics facilitates customized gameplay by forecasting user behavior and dynamically adjusting game mechanics. Through behavioral analysis and preference modeling, the system personalizes content delivery, difficulty settings, and in-game support.

Published by: vikaspatanker

Authors: Mrs. Khatal Kavita, Miss Akanksha Vishwasrao, Miss Nikita Shinde, Miss Apeksha Vishwasrao

Abstract: Managing daily expenses is an important task for people who want to keep track of their finances. The Expense Tracker Web Application is built to make it easier to record, manage, and understand personal financial data. The backend runs on Python Flask, and the front end uses HTML, CSS, and JavaScript to create a user-friendly and responsive interface. The backend supports basic functions like adding, editing, deleting, and viewing expense records, while also ensuring that the data is valid and accurate. It stores information securely in a SQLite database, which allows users to keep and access their financial records easily. The application uses Pandas for handling data and Matplotlib or Plotly for creating visual graphs. This lets users see their spending patterns by category and over time through pie charts and bar or line charts. Additionally, the project focuses on security by cleaning up inputs, checking user data, and optimizing queries for better performance. The system helps users manage daily expenses by organizing data and providing login protection and visual insights to support effective tracking and analysis of spending..

Published by: vikaspatanker

Authors: Prof. Sangeeta Alagi, , Priti Jagdale, Swati More, Vaibhav Prasad

Abstract: The rapid advancement of deep learning technologies has enabled the creation of highly realistic synthetic media, commonly known as deepfakes. These manipulated videos pose serious threats to information integrity, personal privacy, national security, and public trust. This comprehensive literature survey examines the state-of-the-art approaches in deepfake detection, with particular emphasis on methods that combine Convolutional Neural Networks (CNNs) for spatial feature extraction with temporal analysis techniques. We systematically review detection methodologies, benchmark datasets, evaluation metrics, current challenges, and emerging research directions. This survey synthesizes findings from over 50 research papers published between 2018 and 2024, providing insights into the evolution of detection techniques and the ongoing arms race between deepfake generation and detection technologies.

Published by: vikaspatanker

Authors: Aakanksha Raghunath Chaudhari, Sharmistha Sujit Sarkar

Abstract: With the rapid growth of digital communication and online services, network security has become a primary concern for organizations and individuals. Traditional intrusion detection systems (IDS) rely heavily on predefined signatures, making them ineffective against zero-day attacks and unknown threats. To overcome these limitations, AI-based anomaly detection systems have emerged as a powerful approach for identifying unusual patterns in network traffic that may indicate malicious activity. This research introduces NetGuard, an intelligent system that leverages machine learning and deep learning techniques to detect anomalies in network traffic. The system provides real-time threat detection, reduces false alarms, and enhances network resilience against evolving cyber threats.

DOI: https://doi.org/10.5281/zenodo.17570618

 

Published by: vikaspatanker

Authors: Aryan Bhatt, Aryan Verma

Abstract: Artificial Intelligence (AI) methods, specifically Large Language Models (LLMs), are increasingly being employed by developers and students to produce source code. Though helpful, such AI-produced code is problematic in terms of plagiarism, originality, and academic honesty. Hence, differentiating between code written by humans and code generated by AI has become vital for the prevention of plagiarism. This article provides an empirical evaluation of current AI detection tools to determine how well they can detect AI-generated code in educational and coding environments. The findings indicate that most of the tools are ineffective and not generalizable enough to be useful for detecting plagiarism. In order to deal with this problem, we suggest a number of solutions, such as fine-tuning LLMs and machine learning-based classification based on static code metrics and code embeddings obtained from Abstract Syntax Trees (AST). Our top-performing model outperforms current detectors (e.g., GPTSniffer) and gets an F1 score of 82.55. In addition to that, we carry out an ablation study to study the contribution of different source code features to detection accuracy.

DOI:

 

Published by: vikaspatanker