← Back to Archives
VOLUME 13, ISSUE 10, OCTOBER 2025
Application of IOT for Barangay Document Acquisition with Face Recognition Security
Maricon Denber S. Gahisan, Krisseane Marielle B. Sapong, Harbbie Dale A. Elopre, Nordan A. Bañal
Development of Rice Grain Image Classification Model using Artificial Neural Network Architecture
D. R. Solanke, K.D. Chinchkhede, A.B. Manwar
Research Paper on Disaster Management
Prof. Reeta V. Patil, Mr. Rahul K. Kumbhar
AI Powered Security for IoT Networks Ensuring Adaptive Threat Detection Privacy and Resilience
Mohd Abdul Raheem, Moin Uddin Khaja
How to Protecting Children using AI
Prof. Araslan Shaikh, Miss. Sumitra M. Pingale
Crop Disease Classification for Edge Devices: A Quantized MobileNetV2 Approach
Dr. Shilpa Sarvaiya, Pranav Dhole, Ishika Nandwanshi
AI IN CYBER SECURITY
Prof. Vaibhav Chaudhari*, Miss. Pooja Patil
Role Of AI In Precision Agriculture & Smart Farming
Prof. Ms. Chetana Kawale*, Miss.Kalyani.R.Patil
Benefits Of AI In Agriculture Technology
Prof. Mr. Vaibhav Chaudhari*, Mr. Pratham Pradip Patil
Digital Signal Processing
Prof. Arsalan Shaikh*, Mr. Tejas R. Sonawane
Smart attendance using face recognition and web technologies
Mr. Arsalan A Shaikh*, Mr. Lomesh S Yeole
Fuzzy Logic-Based Dual Phase-Shift PWM Strategy for Efficient Bidirectional Wireless Power Transfer in Electric Vehicles
Mutta Murali Tarun, P.Murari
Optimizing EV Charging Infrastructure: A Comparative Study of CC and CC-CV Methods
Pentakota S V V K Naidu, P. Murari
Fuzzy Logic Method for Improving Power Quality in Industrial and Commercial Systems
Kommoju Rohith Kumar, Y. Naveen Kumar
Adaptive Control Strategy for Bidirectional Converters to Improve EV Charging Efficiency
KASIVAJJULA GOPALA KRISHNA ADITYA, K. SATISH KUMAR
An Optimisation Framework for Induction Motor Power in Electric Vehicles Using Fuzzy Logic Control
RAYAVARAPU SAI KIRAN, A. CHAKRADHAR
Stock Price Prediction
Sarthak Agarwal, Devisha Agrawal, Abhishek Singh Rajput, Dr. Golda Dilip
A Multiple Temperature-Control And Alarm System, Based On LM35 Sensor, Using FPGAs and VHDL
Dr Evangelos I. Dimitriadis, Leonidas Dimitriadis
A Full-Bridge Converter of the ZVZCS Type Suitable for MVDC Collection Systems Utilizing Renewable Energy
Chilla Appala Venkata Sai, A. Chakradhar
An Improved Control Strategy for Bidirectional Wireless Power Transfer in Grid-to-Vehicle and Vehicle-to-Home Applications Using a Fuzzy Logic Controller
SURAVARJALA PRADEEP SAI TEJA, CH. VISHNU CHAKRAVARTHI
“GIS – BASED SOLAR POTENTIAL ANALYSIS OF GOVERNMENT BUILDINGS: A CASE STUDY OF MUNICIPAL WARD IN UDAIPUR”, RAJASTHAN, INDIA
Sarvan Kumar, Pooja Kumawat
Performance Of Deep Learning Model For Prediction Of Cryptocurrency Adoption In Nigeria
Sulaiman Umar S.Noma*, Salihu Alhassan, Sadiq Abubakar Zagga and Shamsu Sani
OFF-BOARD ELECTRIC VEHICLE BATTERY CHARGER USING PV ARRAY
RAVADA. GOPAL, U. LAKSHMI
AI-Integrated Smart Manufacturing Systems
Prof. Mr. Arsalan A. Shaikh, Miss. Jayshri Vijay Pawar
Enhancing Credit Card Fraud Detection using The SMOTE and Ensemble Methods
Prof. Arsalan A. Shaikh*, Mr. Mayur Kailas Sapkale
BioGPT-DI: An AI-Powered System for Drug Interaction Prediction and Explainable Clinical Report Generation
Dev Mehta, Sushmita Kundu, Hameed Salihu, Dr. Golda Dilip
GasShield-X: Next-Gen Leakage Detection with Real-Time SMS and SOS Alerts
Bappaditya Mondal, Arnav Moitra, Ankur Banik, Jyotirmoy Pandit*
PlaySmart: A Content-Based Recommender System for Personalized Game Suggestions using Machine Learning
Yogeshwar, Harihara Balan, Praveen Balaji, Dr. Golda Dilip
Data Science and Machine Learning in Student Performance Prediction Using machine learning
Mr. Arsalan A. Shaikh, Shaikh Erfan Gafar
Cyber security in 6G network.
Ms. Chetana Kawale*, Dipesh R Patil
SMS Spam Classifier
Sura Reddy, Sriram K, Brunda G, Dr. Golda Dilip
Unified Payments Interface (UPI) System
Prof. Dr. Dinesh D. Puri, Mr. Rohit Narendra Badgujar
MACHINE LEARNING-BASED NETWORK INTRUSION DETECTION SYSTEM USING THE CSE-CIC-IDS2018 DATASET
Sahithyaa Krishna Kumar, R Sivani, M Jaiaakash, Dr Golda Dilip
Expert Technical Report: Critical Analysis and Strategic Roadmap for Multi-Agent Q-Learning in Heterogeneous Disaster Response
VIGNESH MURALI, T AAKASH, SHARAN S, Dr GOLDA DILIP
Full-Stack Implementation and Evaluation of Abstractive Text Summarization using a Transformer-Based BART Model
MARK OWEN A, PAARIVALAVAN S, SANJAY C, Dr GOLDA DILIP
COMPUTER VISION: TEXT EXTRACTION FROM AN IMAGE
Yokesh Anandan, Christon Davis C, Dr. Golda Dilip
MUSIC GENRE CLASSIFICATION SYSTEM USING NATURAL LANGUAGE PROCESSING
PRANAV H, SARVESH S, HARI SRINIVAS, Dr. Golda Dilip
Machine Learning-Based Plagiarism Detector System
Rakshitha S N, Shreya Sathapathi, Yashvitha J, Dr. Golda Dilip
Enhancing Prediction and Explainability with Machine Learning Using SHAP on OASIS MRI Data Compared to Traditional Diagnosis Methods
Mrinmayi Verma, Neelam Sanjeev Kumar
Advanced Machine Learning for Real-Time Driver Distraction Analysis with Visual Inputs
Ramisetty T M Surya, Sai G, Allam Reddy Charan, Sriram Sanjay S, Neelam Sanjeev Kumar
Hybrid Deep Learning Methodology For Disease Classification In Medical Imaging
Ms. Charulatha RT, M. Adethya, Mani Aadithyaa A.A, Jeevan P, Anto Nawin
PHISHING DETECTION USING CNN AND BiLSTM
Princeton Vishal J, Srinath S, Adithya S, Saran P, Yogesh C, Neelam Sanjeev Kumar
MACHINE LEARNING PREDICTION FOR CROP SELECTION USING A RANDOM FOREST CLASSIFIER
Sreenidhi T, Mithra B V, Lathika M, Kaviya S, Ahana Dinesh, Neelam Sanjeev Kumar
Hybrid PSO-CNN-LSTM Framework for Intelligent DDoS Detection
Dinesh P, Eedpuganti Yagna Sai Harshith, Athithya S A, Raj Pranav Raghavan,G Mohammed Azam, Neelam Sanjeev Kumar
Adaptive Control Strategies for Human-Robot Interaction in Industrial Setting
Nikunj Kaslikar, Aarav Singh, Jeevasree S, Charuhazan B, Neelam Sanjeev Kumar
Abstract
Application of IOT for Barangay Document Acquisition with Face Recognition Security
Maricon Denber S. Gahisan, Krisseane Marielle B. Sapong, Harbbie Dale A. Elopre, Nordan A. Bañal
DOI: 10.17148/IJIREEICE.2025.131001
Abstract: In today’s digital age, the traditional method in acquiring the documents in government offices often involve lengthy queues, paperwork, and time-consuming processes. This manual approach can be inefficient, prone to errors, and frustrating for both residents and government workers. To address this challenge and enhance the overall efficiency of the barangay services, the development and implementation of an automated document acquisition system provided a promising solution. This developed design utilized the internet and databases to request for a specific type of document with the face recognition security as the only way for getting the request in the absence of the barangay official or personnel. It also featured an email notification to alert the user should their document request be approved or disapproved. The developed system was tested based on its performance with an average turnaround time of 73.55 seconds from the time the user was able to login up to the printing of document resulting to significant improvement on the processing time. As to the user experience it was found out that the system is easy to navigate with an acceptable user interface response time within 2-6.3 seconds making it a user-friendly device.
Keywords: Document Acquisition System, Barangay Documents Acquisition, IOT, Raspberry Pi
Keywords: Document Acquisition System, Barangay Documents Acquisition, IOT, Raspberry Pi
Abstract
Development of Rice Grain Image Classification Model using Artificial Neural Network Architecture
D. R. Solanke, K.D. Chinchkhede, A.B. Manwar
DOI: 10.17148/IJIREEICE.2025.131002
Abstract: Rice is one of the most widely consumed staple crops globally, and its quality assessment plays a crucial role in food safety, trade, and agricultural productivity. Traditional methods of rice grain classification are largely manual, time-consuming, and prone to human error. In this study, we propose the development of an automated Rice Grain Image Classification Model using Artificial Neural Network (ANN) architecture. A dataset comprising high-resolution images of various rice grain types was collected and pre-processed, including resizing, normalization, and feature extraction. The ANN model was designed to capture subtle morphological and textural differences among grain categories, enabling accurate classification. Experimental results demonstrate that the proposed model achieves high accuracy, robust generalization, and reliable performance across different grain classes. The study highlights the potential of ANN-based approaches for automating rice quality assessment, reducing human intervention, and improving efficiency in post- harvest processing and market evaluation. The proposed framework can be extended to other agricultural products, supporting intelligent and data-driven quality management in the agro-food sector.
Keywords: ANN, Accuracy, Classification, Cross-entropy, Machine learning, precision.
Keywords: ANN, Accuracy, Classification, Cross-entropy, Machine learning, precision.
Abstract
Research Paper on Disaster Management
Prof. Reeta V. Patil, Mr. Rahul K. Kumbhar
DOI: 10.17148/IJIREEICE.2025.131003
Abstract: Disasters—whether natural or man-made—pose significant threats to human life, infrastructure, and the environment. The increasing frequency and intensity of disasters due to climate change, urbanization, and industrialization highlight the urgent need for effective disaster management strategies. This research paper explores disaster management as a multidisciplinary field encompassing prevention, preparedness, response, and recovery. The paper reviews existing approaches, analyzes modern technological interventions, and proposes an integrated model combining traditional strategies with advanced tools such as Geographic Information Systems (GIS), Artificial Intelligence (AI), and early warning systems. The study concludes that collaborative governance, community participation, and technological innovation are key to reducing disaster risk and enhancing resilience.
1
1
Abstract
AI Powered Security for IoT Networks Ensuring Adaptive Threat Detection Privacy and Resilience
Mohd Abdul Raheem, Moin Uddin Khaja
DOI: 10.17148/IJIREEICE.2025.131004
Abstract: The proliferation of IoT devices across industries has revolutionized efficiency but created an expansive, complex attack surface characterized by heterogeneous devices, weak protocols, and low physical security. Conventional security solutions are often ineffective due to IoT’s resource constraints and unique latency and scalability needs. Artificial Intelligence approaches—spanning deep learning, federated, and edge-based frameworks—address these gaps through adaptive, autonomous, and privacy-aware threat detection using real-time telemetry and behavioural analytics. Techniques such as intrusion detection, device fingerprinting, and anomaly detection enable timely response against known and novel threats. This review surveys leading AI strategies for IoT security, explores dataset benchmarks, adversarial resilience, resource allocation, explainable AI, and privacy safeguards. Ongoing challenges include defending against advanced persistent threats, ensuring robust operation across diverse environments, optimizing efficiency, and providing standardized datasets. The findings advise stakeholders on building scalable, trustworthy, and resilient AI- powered IoT security systems.
Keywords: Edge AI Security, IoT Anomaly Detection, Device Fingerprinting, Botnet Detection, Bot-IoT Dataset, N- BaIoT Dataset, Privacy-Preserving Machine Learning, Adversarial Machine Learning for IoT, Explainable AI (XAI) for IoT, Hybrid Edge-Cloud Security Architecture, IoT Threat Modelling, Distributed Intrusion Detection, Secure Federated Aggregation, and real-time threat mitigation for IoT.
Keywords: Edge AI Security, IoT Anomaly Detection, Device Fingerprinting, Botnet Detection, Bot-IoT Dataset, N- BaIoT Dataset, Privacy-Preserving Machine Learning, Adversarial Machine Learning for IoT, Explainable AI (XAI) for IoT, Hybrid Edge-Cloud Security Architecture, IoT Threat Modelling, Distributed Intrusion Detection, Secure Federated Aggregation, and real-time threat mitigation for IoT.
Abstract
How to Protecting Children using AI
Prof. Araslan Shaikh, Miss. Sumitra M. Pingale
DOI: 10.17148/IJIREEICE.2025.131005
Abstract: Children today confront a growing number of threats in both digital and physical surroundings, including exposure to dangerous content, cyberbullying, and exploitation. Traditional defenses are often inadequate due to the magnitude and complexity of modern threats. This study investigates how scalable, proactive, and intelligent protection solutions provided by artificial intelligence (AI) might enhance kid safety. The study's objectives are to identify AI-based child protection techniques, evaluate their effectiveness, and raise awareness of privacy and ethical concerns. The literature review, system architecture design, and analysis of current AI-driven child protection tools are all combined in a mixed-methods approach. We examine methods such as natural language processing for content filtering, computer vision for online interaction monitoring, and machine learning for behavior prediction.
Abstract
Crop Disease Classification for Edge Devices: A Quantized MobileNetV2 Approach
Dr. Shilpa Sarvaiya, Pranav Dhole, Ishika Nandwanshi
DOI: 10.17148/IJIREEICE.2025.131006
Abstract: Agriculture plays a critical role in global food security, yet crop diseases continue to cause significant eco- nomic losses worldwide. Traditional deep learning models, while achieving high accuracy in disease detection, face substantial limitations when deployed for real-time, in-field applications due to their computational complexity and large memory requirements. This research addresses these challenges by developing a lightweight, quantized model specifi- cally designed for edge device deployment. We propose applying model compression techniques through quantization on MobileNetV2, a lightweight neural network architecture, to create an efficient model suitable for resource-constrained environments. Our methodology involves comprehensive comparison of Post-Training Quantization (PTQ) and Dynamic Range Quantization (DRQ) techniques applied to rice leaf disease classification. The results demonstrate a significant reduction in model size from approximately 9 MB to 2.5 MB while maintaining acceptable accuracy levels. The DRQ model achieved 92.23% accuracy with an F1-score of 0.9212, compared to the original model’s 94% accuracy, repre- senting a minimal 1.77% accuracy trade-off for a 72% size reduction. These findings highlight the practical viability of quantized models for automated disease detection systems in precision agriculture, enabling real-time deployment on smartphones and embedded devices for farmers in remote locations.
Keywords: Crop disease classification, edge computing, model quantization, MobileNetV2, precision agriculture, deep learning compression.
Keywords: Crop disease classification, edge computing, model quantization, MobileNetV2, precision agriculture, deep learning compression.
Abstract
AI IN CYBER SECURITY
Prof. Vaibhav Chaudhari*, Miss. Pooja Patil
DOI: 10.17148/IJIREEICE.2025.131007
Abstract: The integration of Artificial Intelligence (AI) into cyber security has revolutionized the way organizations detect, prevent, and respond to cyber threats. AI technologies, including machine learning, deep learning, and natural language processing, enable systems to analyze vast amounts of data in real time, identify patterns, and predict potential security breaches with high accuracy. These intelligent systems can autonomously detect anomalies, identify zero-day vulnerabilities, and respond to threats faster than traditional methods. Moreover, AI enhances threat intelligence, automates routine security tasks, and supports adaptive defense mechanisms, reducing the burden on human analysts. However, the use of AI in cybersecurity also introduces new challenges, such as adversarial attacks, data privacy concerns, and the potential misuse of AI by malicious actors. This paper explores the current applications, benefits, limitations, and future prospects of AI in cyber security, highlighting its critical role in building resilient digital infrastructures.
Abstract
Role Of AI In Precision Agriculture & Smart Farming
Prof. Ms. Chetana Kawale*, Miss.Kalyani.R.Patil
DOI: 10.17148/IJIREEICE.2025.131008
Abstract: Agriculture faces increasing challenges from population growth, climate change, resource scarcity, and the need for sustainable food production. Traditional farming practices often rely on generalized methods that lead to inefficiencies in crop management and resource utilization. This paper explores the role of Artificial Intelligence (AI) in precision agriculture and smart farming as a solution to these challenges. The main objective is to assess how AI-driven techniques can optimize agricultural operations, enhance productivity, and ensure environmental sustainability. A comprehensive review methodology was employed, drawing upon recent studies, case analyses, and technological applications such as machine learning, computer vision, robotics, and IoT-based data collection. Findings indicate that AI enables accurate crop monitoring, predictive yield estimation, soil and water management, and early detection of pests and diseases. Moreover, AI-powered autonomous systems improve efficiency in planting, irrigation, and harvesting, reducing labor dependency and operational costs. The study highlights that while AI adoption significantly boosts decision-making and resource optimization, challenges such as high initial costs, lack of rural digital infrastructure, and data privacy concerns remain barriers. In conclusion, AI has emerged as a transformative tool for precision agriculture and smart farming, offering promising pathways toward food security, climate resilience, and sustainable agricultural development.
Abstract
Digital Signal Processing
Prof. Arsalan Shaikh*, Mr. Tejas R. Sonawane
DOI: 10.17148/IJIREEICE.2025.131010
Abstract: Digital signal processing (DSP), which deals with the digital representation, processing, and analysis of signals, is a crucial field in modern engineering. This paper's primary goal is to increase the precision and effectiveness of signal processing techniques in order to meet the growing demands of applications such as control systems, biomedical engineering, audio and image processing, and telecommunications. Investigating basic DSP concepts, including sampling, quantization, discrete-time signal encoding, and the use of Fourier and Z-transforms, is the main objective. Hardware and algorithm implementations of these concepts are also examined. The methodology includes theoretical study, simulation, and experimental validation to demonstrate the effectiveness of several DSP algorithms for noise reduction, signal filtering, and feature extraction.
Abstract
Smart attendance using face recognition and web technologies
Mr. Arsalan A Shaikh*, Mr. Lomesh S Yeole
DOI: 10.17148/IJIREEICE.2025.131011
Abstract: Attendance management is critical for institutions, but traditional procedures like manual roll calls and sign-in sheets usually face problems like human error, time consumption, proxy attendance ("buddy punching"), and a lack of real-time data. Face recognition technology, integrated with web platforms, offers a workable solution with its automated, contactless, and accurate identification process. By providing a secure, efficient, and easily accessible system for administrators to monitor presence, this smart attendance system has the potential to revolutionize daily administrative workflows. Proxy attendance, data entry errors, and significant administrative overhead are some of the problems with traditional paper-based attendance that can be resolved by integrating computer vision into management systems. With this smart, web-based attendance system, registered individuals can have their attendance marked automatically and instantly using a simple webcam. This approach makes use of powerful machine learning models and image processing, two of computer vision's built-in strengths, to guarantee the integrity of the attendance records. Additionally, by facilitating a touchless and rapid process, the system can improve operational efficiency and reduce disruptions. Attendance marking is made secure, transparent, and accurate with this system built on Python and OpenCV. Because it is a web application powered by a machine learning back-end, administrators can view live data and generate reports without manual compilation. Because facial encodings and timestamps are stored in a secure database, it is impossible for unauthorized individuals to tamper with or alter the records. The automated, transparent, and reliable attendance process is guaranteed by the use of facial recognition algorithms. With the potential to increase data accuracy and confidence in administrative records, face recognition technology and the deployment of a smart system provide a dependable and affordable solution for efficient and trustworthy attendance management.
Keywords: Face Recognition, Smart Attendance, Web Technologies, Computer Vision, Biometric System.
Keywords: Face Recognition, Smart Attendance, Web Technologies, Computer Vision, Biometric System.
Abstract
Fuzzy Logic-Based Dual Phase-Shift PWM Strategy for Efficient Bidirectional Wireless Power Transfer in Electric Vehicles
Mutta Murali Tarun, P.Murari
DOI: 10.17148/IJIREEICE.2025.131012
Abstract: Bidirectional Wireless Power Transfer (BWPT) removes the need for physical intervention to facilitate smooth Grid-to-Vehicle (G2V) and Vehicle-to-Grid (V2G) operations; nevertheless, it still has drawbacks, including efficiency issues, power factor management, and low transfer rates. In order to enhance Power Factor Correction (PFC) in BWPT systems, this study suggests a Dual Phase Shift Pulse Width Modulation (DPS-PWM) technique. To improve the dynamic reaction and optimise switching choices in real-time depending on input changes and system behaviour, a Fuzzy Logic Controller (FLC) is also incorporated.
Simulation and experimental setups running at 85 kHz and 3.7 kW are used to assess the suggested intelligent control approach. The BWPT system is modelled and simulated using MATLAB/Simulink, which enables thorough performance study under various operating circumstances. The results show significant gains in Total Harmonic Distortion (THD) and overall system robustness, with a power transfer efficiency of 90.1% in the experiment and 94.4% in the simulation. Fuzzy logic integration has great promise for effective and adaptive control in next-generation BWPT systems for electric cars.
Simulation and experimental setups running at 85 kHz and 3.7 kW are used to assess the suggested intelligent control approach. The BWPT system is modelled and simulated using MATLAB/Simulink, which enables thorough performance study under various operating circumstances. The results show significant gains in Total Harmonic Distortion (THD) and overall system robustness, with a power transfer efficiency of 90.1% in the experiment and 94.4% in the simulation. Fuzzy logic integration has great promise for effective and adaptive control in next-generation BWPT systems for electric cars.
Abstract
Optimizing EV Charging Infrastructure: A Comparative Study of CC and CC-CV Methods
Pentakota S V V K Naidu, P. Murari
DOI: 10.17148/IJIREEICE.2025.131013
Abstract: The development of Electric Vehicles (EV) has seen lithium batteries emerge as the primary energy source in recent years, owing to their peak energy density, enhanced power density, and extended lifespan. Rapid and efficient battery charging is essential for electric vehicles. While it takes only a few minutes to refuel petrol-powered vehicles, an EV requires 4-6 hours for a full charge, depending on the C-rate. This paper discusses the modeling and simulation of a multi-current charging method designed for a two-wheeled electric vehicle. The proposed approach employs a closed- loop control system to regulate the charging current via a buck converter power conditioning circuit. To assess the effectiveness of the proposed charging method, the circuit is simulated within the MATLAB/Simulink environment, and the results are compared to those obtained from the constant current charging (CC) method and the Constant Current- Constant Voltage (CC-CV) charging method.
Abstract
Fuzzy Logic Method for Improving Power Quality in Industrial and Commercial Systems
Kommoju Rohith Kumar, Y. Naveen Kumar
DOI: 10.17148/IJIREEICE.2025.131014
Abstract: This document presents an innovative control strategy based on fuzzy logic for multifunctional grid-tied inverters utilized in industrial and commercial power systems. The primary objective is to mitigate instantaneous power oscillations and improve overall power quality. The proposed strategy incorporates the principles of Conservative Power Theory (CPT), which facilitates the direct extraction and analysis of oscillating power components within the ABC frame. Additionally, fuzzy logic contributes adaptive intelligence to effectively manage uncertainties and nonlinearities associated with complex load conditions. To assess the efficacy of the proposed fuzzy control method, comprehensive simulations are performed on a three-phase multifunctional grid-tied inverter that is fitted with an LCL filter in a laboratory-scale prototype setting. The power system model encompasses a variety of load profiles, including linear, nonlinear, unbalanced loads, and a three-phase induction motor with capacitive compensation. The simulation outcomes reveal that the fuzzy logic-enhanced control strategy not only optimizes active power dispatch by the inverter but also guarantees constant torque operation of the induction motor, thereby enhancing power quality across the entire industrial/commercial system. Moreover, the strategy is experimentally validated using a 3.6 kVA inverter prototype. The experimental findings corroborate the successful compensation of the targeted oscillating power components, highlighting the practical viability and robustness of the fuzzy logic-based approach in real-world scenarios.
Abstract
Adaptive Control Strategy for Bidirectional Converters to Improve EV Charging Efficiency
KASIVAJJULA GOPALA KRISHNA ADITYA, K. SATISH KUMAR
DOI: 10.17148/IJIREEICE.2025.131015
Abstract: This research introduces an ANN-based bidirectional power converter designed to enhance EV charging control networks. The proposed system efficiently manages the charging and discharging of electric vehicles by employing an ANN controller that dynamically regulates the power flow in bidirectional converters. By leveraging the adaptive capabilities of the ANN, the converter adjusts in real time to changing charging requirements. This approach maximizes the stability and efficiency of energy transfer across various load conditions. Unlike traditional controllers, this method reduces response time while simultaneously enhancing power quality, facilitating quicker and more efficient charging. Experimental results indicate that the system is capable of charging different electric vehicle batteries with varying degrees of efficiency and precision. An ANN-based bidirectional converter has the potential to make electric car chargers smaller, lighter, and more affordable, while also providing a scalable solution to accommodate the growing electric vehicle network.
Abstract
An Optimisation Framework for Induction Motor Power in Electric Vehicles Using Fuzzy Logic Control
RAYAVARAPU SAI KIRAN, A. CHAKRADHAR
DOI: 10.17148/IJIREEICE.2025.131016
Abstract: Energy efficiency plays a vital role in electric and hybrid vehicles (EVs) due to their limited energy storage capacity. Besides its excellent stability and low cost, minimizing losses enhances the efficiency of the induction motor. Furthermore, when operating below its maximum load, it may consume more power than necessary for its tasks. This study proposes a fuzzy logic control (FLC)-based method for applications in electric vehicles. The FLC controller is capable of conserving additional power and improving the initial current amplitude. The performance of the control was validated through simulation using the MATLAB/SIMULINK software program. The simulation results indicate a swift rejection of disturbances impacting the system and demonstrate superior performance outcomes in the time-domain response compared to the conventional proportional integral derivative controller. Consequently, the core losses of the induction motor are significantly reduced, thereby enhancing the efficiency of the driving system.
Abstract
Stock Price Prediction
Sarthak Agarwal, Devisha Agrawal, Abhishek Singh Rajput, Dr. Golda Dilip
DOI: 10.17148/IJIREEICE.2025.131017
Abstract: Stock markets are volatile and influenced by many factors, making price prediction difficult. This project presents a web app that predicts stock prices using a trained Long Short-Term Memory (LSTM) model. The system fetches historical data from Yahoo Finance, preprocesses it, and predicts future prices for both U.S. and Indian markets. The LSTM model captures time-based trends effectively and provides accurate forecasts. The app also includes visualization tools and performance metrics for better analysis
Keywords: LSTM, Stock Price Prediction, Machine Learning, Deep Learning, Time Series Forecasting
Keywords: LSTM, Stock Price Prediction, Machine Learning, Deep Learning, Time Series Forecasting
Abstract
A Multiple Temperature-Control And Alarm System, Based On LM35 Sensor, Using FPGAs and VHDL
Dr Evangelos I. Dimitriadis, Leonidas Dimitriadis
DOI: 10.17148/IJIREEICE.2025.131018
Abstract: A multiple temperature-control system, based on LM35 sensor, FPGAs and VHDL, is presented here. The system is capable of providing a series of controls and subsequently activate respective alarm systems. It can simulta- neously monitor three basic temperature-related parameters. The first is temperature range values of specific area or a human. Blue, green and red LEDs light up, to present temperature lowering below lower limit value, temperature re- maining within set values or exceeding upper set value, respectively. If temperature is out of limits buzzer also sounds. Second basic parameter monitored here, is temperature rising or lowering rate within specific time set values and sub- sequent activation of center yellow LED and half right of board LEDs or center white LED and half left of board LEDs, respectively. Finally the third basic parameter monitored with our system, is temperature remaining above upper criti- cal set value or below lower critical set value for a specific time period, thus activating breadboard’s right yellow or white LEDs, respectively. LM35 temperature sensor used here and its analog voltage values act as input to FPGA’s ADC unit and converted temperature values are presented to seven-segment displays. All LED systems activated here, correspond to related external control systems which are activated on a case-by-case basis. The system uses DE10-Lite FPGA board and taking into account that specific time periods, as well as upper and lower temperature limits and tem- perature rising or lowering rates can be set to a variety of values, gives our system the ability of implementation in a wide range of applications such as patient, room, laboratory, industrial or external environment monitoring.
Keywords: LM35 sensor, Temperature, FPGA, VHDL, Buzzer, LEDs.
Keywords: LM35 sensor, Temperature, FPGA, VHDL, Buzzer, LEDs.
Abstract
A Full-Bridge Converter of the ZVZCS Type Suitable for MVDC Collection Systems Utilizing Renewable Energy
Chilla Appala Venkata Sai, A. Chakradhar
DOI: 10.17148/IJIREEICE.2025.131019
Abstract: This research presents a full-bridge DC-DC converter that utilizes zero-voltage zero-current switching (ZVZCS), structured around a dual-transformer configuration with two output filter capacitors. The converter is engineered for seamless integration with DC collection systems in medium-voltage renewable energy applications, where stability, adaptability, and efficiency are of utmost importance. The primary switches function under a pulse width modulation (PWM) strategy with a fixed duty cycle to control voltage and power output. Under full load conditions, the main full-bridge circuit provides the majority of the power and achieves zero-voltage switching (ZVS) due to the meticulous design of the main transformer’s turn ratio, which significantly reduces switching losses. To support this, an auxiliary circuit implements ZVZCS and manages a minor portion of the power. Efficiency can be further enhanced by fine-tuning the auxiliary transformer's turns ratio, which balances power contribution while minimizing conduction losses. In addition to traditional control techniques, a fuzzy logic controller is integrated to improve real-time adaptability. The fuzzy controller dynamically modifies control signals in response to fluctuating input voltage, load conditions, and switching timing, enabling the converter to sustain soft-switching conditions even amidst transient disturbances or parameter variations. It supersedes inflexible threshold-based decision-making. Utilizing linguistic principles, this approach facilitates a more intelligent management of system nonlinearity and uncertainty. The design features, the definition of the fuzzy rule base, and the optimization strategies for the suggested converter are thoroughly detailed. In order to confirm the efficacy of the proposed method and to validate the simulation outcomes, a prototype rated at 200 V / 2 kv / 3 kw has been constructed and evaluated. The findings indicate enhanced switching performance, improved thermal management, and increased overall efficiency as a result of the integrated application of ZVZCS and fuzzy logic control.
Abstract
An Improved Control Strategy for Bidirectional Wireless Power Transfer in Grid-to-Vehicle and Vehicle-to-Home Applications Using a Fuzzy Logic Controller
SURAVARJALA PRADEEP SAI TEJA, CH. VISHNU CHAKRAVARTHI
DOI: 10.17148/IJIREEICE.2025.131020
Abstract: The progress in the charging strategies for electric vehicles is expected to have significant impacts on the electric grid in the near future. Electric vehicle battery chargers are designed to facilitate bidirectional power transfer in accordance with the vehicle-to-grid concept, thereby providing essential services to both the distribution grid and the domestic grid of the vehicle owner. Wireless power transfer battery chargers present a safer and more user-friendly option for individuals who may lack confidence in handling technological devices. Bidirectional wireless power transfer chargers that support vehicle-to-grid services represent a natural progression of the previously mentioned concepts. This paper addresses the development of a control strategy for such a battery charger, concentrating on the requirements of the power conversion stages necessary for the operation of a charger designed for vehicle-to-home functionality. Initially, the division of the control strategy into two distinct levels is discussed, followed by an introduction to the interaction between the algorithms at the internal and external levels. In the implementation of the control algorithms, the decision was made to design the controllers with simplicity in mind. This approach allowed for the adoption of well-established techniques within the scientific community for their design, while also minimizing the computational resources required for their execution. Despite the straightforward nature of the controllers, the introduction and management of interactions among the various algorithms resulted in the formulation of a comprehensive control strategy that simultaneously adheres to the voltage and current limits imposed by the grid and the battery, while also preventing the maximum operating conditions of the static converters that comprise the system from being exceeded. The algorithms and their corresponding controllers are developed sequentially in the continuous time domain, utilizing techniques grounded in the analysis of Bode diagrams of the transfer functions integral to the system's operation. In the design of the controllers, the implications of their subsequent effects are also taken into account.
Abstract
“GIS – BASED SOLAR POTENTIAL ANALYSIS OF GOVERNMENT BUILDINGS: A CASE STUDY OF MUNICIPAL WARD IN UDAIPUR”, RAJASTHAN, INDIA
Sarvan Kumar, Pooja Kumawat
DOI: 10.17148/IJIREEICE.2025.131021
Abstract: The increasing demand for renewable energy solutions has emphasized the need for efficient rooftop solar photovoltaic (PV) system assessment in urban areas. This study integrates geographic Information System (GIS) and Remote Sensing techniques to evaluate the solar potential of government buildings in a municipal ward pf Udaipur. The methodology includes ward boundary and rooftop boundary digitization using Google Earth Imagery & the height of buildings was measured in Ruler tool, followed by the Extraction of Digital Surface Model (DSM) data from NASA sources. Solar insolation analysis is conducted in SAGA GIS to identify the most suitable rooftop for solar panel installation. Additionally existing solar panel are mapped, and optimal locations for new installations are proposed using Helioscope software, considering factors such as panel tilt, azimuth angle, and energy generation capacity. The study provides a comprehensive framework for urban solar suitability mapping.
Keywords: GIS, Remote Sensing, Digital Surface Model, Rooftop Photovoltaic, Helioscope
Keywords: GIS, Remote Sensing, Digital Surface Model, Rooftop Photovoltaic, Helioscope
Abstract
Performance Of Deep Learning Model For Prediction Of Cryptocurrency Adoption In Nigeria
Sulaiman Umar S.Noma*, Salihu Alhassan, Sadiq Abubakar Zagga and Shamsu Sani
DOI: 10.17148/IJIREEICE.2025.131022
Abstract: The use of cryptocurrencies is rapidly increasing in developing countries as governments and financial institutions become more aware of them. With millions of people actively transacting in digital assets, Nigeria is a global leader in the adoption of cryptocurrencies. Our knowledge of the socioeconomic elements influencing adoption is still lacking, despite this. Gaps remain in understanding the socio-economic drivers of adoption. This study explores cryptocurrency adoption in Nigeria by using Convolutional Neural Networks (CNN) for predictive analytics. We develop and evaluate a CNN model using adoption-related datasets to classify and predict adoption trends. The CNN model was evaluated on five (5) performance evaluation metrics and achieved 92% accuracy, 90% precision, an 86% recall score, an 86% F1 score and a 95% ROC-AUC. Therefore, results indicate that CNN can effectively capture nonlinear relationships in adoption patterns, outperforming traditional machine learning models in accuracy and generalisation. The study revealed that Convolutional Neural Networks (CNN) can accurately estimate and forecast Nigeria's adoption of cryptocurrency and provides insights for policymakers, financial institutions and technology innovators.
Keywords: Deep learning, Cryptocurrency, Convolutional Neural Network (CNN), Adoption, Nigeria.
Keywords: Deep learning, Cryptocurrency, Convolutional Neural Network (CNN), Adoption, Nigeria.
Abstract
OFF-BOARD ELECTRIC VEHICLE BATTERY CHARGER USING PV ARRAY
RAVADA. GOPAL, U. LAKSHMI
DOI: 10.17148/IJIREEICE.2025.131023
Abstract: In the past decade, the automobile sector has experienced significant growth due to the advancement of electric vehicles (EVs). The battery charging system is crucial for the progress of EVs. Charging EV batteries from the grid increases the demand on the load. Consequently, this study proposes a photovoltaic (PV) array-based off-board EV battery charging system. Regardless of solar irradiance levels, the EV battery must be charged consistently, which is accomplished by incorporating a backup battery bank alongside the PV array. By utilizing a sepic converter and a three- phase bidirectional DC–DC converter, the proposed system can charge the EV battery during both sunny and non-sunny periods. During peak sunlight hours, the backup battery charges in conjunction with the EV battery, while during non- sunny hours, the backup battery facilitates the charging of the EV battery. The proposed charging system has been simulated using Simulink within the MATLAB software, and an experimental prototype has been constructed and tested in the laboratory, with the results presented in this study.
Abstract
AI-Integrated Smart Manufacturing Systems
Prof. Mr. Arsalan A. Shaikh, Miss. Jayshri Vijay Pawar
DOI: 10.17148/IJIREEICE.2025.131024
Abstract: Research Problem: Traditional manufacturing systems often struggle with inefficiencies, equipment downtime, and a lack of real-time adaptability. These issues stem from rigid automation that lacks intelligent decision-making capabilities. Objectives: • To explore how AI enhances smart manufacturing systems through predictive maintenance, quality control, and real-time optimization.
• To evaluate AI-enabled manufacturing performance metrics.
• To assess the potential impact of these technologies on operational efficiency and industry transformation. Methods: This paper uses a combination of simulation-based experiments and case study analysis in automotive and electronics sectors. It incorporates deep learning models for fault detection and predictive maintenance. Key Findings: The AI-integrated manufacturing system demonstrated a 30% reduction in unplanned downtime, 20% improvement in product quality, and a 25% increase in throughput compared to traditional systems.
Conclusion: The results underscore AI’s transformative potential in creating intelligent, selfoptimizing factories aligned with Industry 4.0 and paving the way for Industry 5.0.
• To evaluate AI-enabled manufacturing performance metrics.
• To assess the potential impact of these technologies on operational efficiency and industry transformation. Methods: This paper uses a combination of simulation-based experiments and case study analysis in automotive and electronics sectors. It incorporates deep learning models for fault detection and predictive maintenance. Key Findings: The AI-integrated manufacturing system demonstrated a 30% reduction in unplanned downtime, 20% improvement in product quality, and a 25% increase in throughput compared to traditional systems.
Conclusion: The results underscore AI’s transformative potential in creating intelligent, selfoptimizing factories aligned with Industry 4.0 and paving the way for Industry 5.0.
Abstract
Enhancing Credit Card Fraud Detection using The SMOTE and Ensemble Methods
Prof. Arsalan A. Shaikh*, Mr. Mayur Kailas Sapkale
DOI: 10.17148/IJIREEICE.2025.131025
Abstract: In the current days, with, the usage of credit cards has increased radically due to its varied benefits. The mode of payment through credit card has made people’s life easy for both online and ordinary purchases and thus widespread. This enormous usage of credit card leads to different frauds. Due to the rise and rapid growth of E-Commerce, use of credit cards for online purchases has dramatically increased and it caused an explosion in the credit card fraud. As credit card becomes the most popular mode of payment for both online as well as regular purchase, cases of fraud associated with it are also rising. In real life, fraudulent transactions are scattered with genuine transactions and simple pattern matching techniques are not often sufficient to detect those frauds accurately. This system seeks to investigate the current debate regarding the credit fraud in the banking sector and vulnerabilities in online banking and to study some possible remedial actions to detect and prevent credit fraud. The system reveals lots of channels of fraud in online banking which are increasing day by day. These kinds of fraud are the main barriers for the e-business in the banking sector. This system also gives the details of a survey of various techniques used in credit card fraud detection mechanisms and evaluates each methodology based on certain design criteria.
Keywords: Fraud detection, credit card, SMOTE and Ensemble Method, E-Commerce.
Keywords: Fraud detection, credit card, SMOTE and Ensemble Method, E-Commerce.
Abstract
BioGPT-DI: An AI-Powered System for Drug Interaction Prediction and Explainable Clinical Report Generation
Dev Mehta, Sushmita Kundu, Hameed Salihu, Dr. Golda Dilip
DOI: 10.17148/IJIREEICE.2025.131026
Abstract: The increasing complexity of polypharmacy presents a significant challenge to patient safety, with adverse drug reactions (ADRs) stemming from drug-drug interactions (DDIs) representing a major cause of morbidity and mortality. Traditional DDI checking systems, which often rely on static databases, lack the contextual nuance required for effective clinical decision-making. This paper introduces Bigot-DI, an intelligent, networked application designed to predict and explain DDIs using a state-of-the-art, two-engine AI architecture. The system leverages a fine-tuned BioBERT model for high-accuracy DDI classification and a generative BioGPT model to produce real-time, audience- specific clinical summaries for both healthcare professionals and patients. By analyzing a drug pair, the system can predict the interaction type and generate a detailed report on its potential effects and mechanisms, transforming a simple query into an actionable clinical insight. This paper provides a complete blueprint for the development and deployment of this serverless application, from the fine-tuning of its biomedical language models to the design of its scalable backend API and modern frontend interface. Future work will focus on integrating diverse data sources, such as real-world evidence from the TWOSIDES dataset, to further enhance predictive accuracy and enrich the clinical reports.
Keywords: Biomedical NLP, Drug-Drug Interaction (DDI), Generative AI, Explainable AI, Clinical Decision Support, BioBERT, BioGPT, Health AI, Bioinformatics.
Keywords: Biomedical NLP, Drug-Drug Interaction (DDI), Generative AI, Explainable AI, Clinical Decision Support, BioBERT, BioGPT, Health AI, Bioinformatics.
Abstract
GasShield-X: Next-Gen Leakage Detection with Real-Time SMS and SOS Alerts
Bappaditya Mondal, Arnav Moitra, Ankur Banik, Jyotirmoy Pandit*
DOI: 10.17148/IJIREEICE.2025.131027
Abstract: Liquefied Petroleum Gas (LPG) is a widely used energy source for domestic, commercial, and industrial applications; however, accidental leakages pose serious threats such as explosions, fire hazards, and health risks resulting from gas inhalation or suffocation. To address these concerns, this study proposes a smart LPG leakage detection and alert system integrated with an SOS emergency response feature. The system employs an MQ-2 smoke and gas sensor to continuously monitor LPG concentration levels in the environment. When the gas concentration exceeds a predetermined safety threshold, the Arduino UNO R3 CH340G microcontroller processes the signal and triggers an SMS alert through the SIM800L GPRS GSM module, immediately notifying users about the leakage. If no action is taken within a specified time, the system automatically initiates an SOS call to the nearest fire station, transmitting the GPS location for a rapid emergency response. Designed to be cost-effective, compact, and easy to install, the system enhances safety in residential and industrial environments by providing real-time alerts and automated communication during critical situations. The proposed work presents the design, implementation, and performance evaluation of the system, demonstrating its reliability, efficiency, and potential for large-scale deployment in safety-critical applications.
Keywords: LPG, GSM technologies, Gas Sensor, Arduino, SOS Alert System, Real-Time Monitoring.
Keywords: LPG, GSM technologies, Gas Sensor, Arduino, SOS Alert System, Real-Time Monitoring.
Abstract
PlaySmart: A Content-Based Recommender System for Personalized Game Suggestions using Machine Learning
Yogeshwar, Harihara Balan, Praveen Balaji, Dr. Golda Dilip
DOI: 10.17148/IJIREEICE.2025.131028
Abstract: With the continuous growth of the gaming industry, players often face difficulty discovering new games that match their interests. The vast collection of games on digital platforms such as Steam can overwhelm users, making personalized recommendations crucial for enhancing user experience. This paper proposes PlaySmart, a content-based recommender system that leverages machine learning techniques to suggest games based on their similarity to titles previously enjoyed by users. Using the Steam dataset, game metadata such as genres, developers, tags, and descriptions are processed using the TF-IDF (Term Frequency–Inverse Document Frequency) vectorization technique and cosine similarity to compute recommendations. The proposed model provides relevant, accurate, and personalized results while maintaining simplicity and interpretability.The proposed system demonstrates how content-based filtering can effectively personalize recommendations while maintaining simplicity, scalability, and transparency key factors in modern recommender systems for digital entertainment platforms.
Keywords: Machine Learning, Recommender System, Content-Based Filtering, TF-IDF, Steam Dataset, Cosine Similarity
Keywords: Machine Learning, Recommender System, Content-Based Filtering, TF-IDF, Steam Dataset, Cosine Similarity
Abstract
Data Science and Machine Learning in Student Performance Prediction Using machine learning
Mr. Arsalan A. Shaikh, Shaikh Erfan Gafar
DOI: 10.17148/IJIREEICE.2025.131029
Abstract: Education in the modern era is increasingly shaped by data-driven technologies that transform traditional learning systems into intelligent and adaptive environments. Predicting student performance has become one of the most significant research areas in educational data mining (EDM) and learning analytics (LA). Accurate prediction enables educational institutions to identify at-risk students early, plan interventions, and promote personalized learning experiences.
This research explores how Data Science and Machine Learning (ML) can be applied to predict student academic performance using structured datasets. It highlights algorithms such as Linear Regression, Decision Tree, Random Forest, Support Vector Machine (SVM), and Artificial Neural Network (ANN), demonstrating their potential to analyze educational data and forecast learning outcomes. The study employs Python-based tools such as Scikit-learn, Pandas, and NumPy for model training, testing, and evaluation.
A complete ML pipeline is designed — including data collection, preprocessing, feature selection, model development, and performance evaluation — to predict student grades and categorize learners into performance classes such as High, Medium, and Low. Performance metrics like accuracy, precision, recall, and F1-score are used to evaluate model effectiveness.
The research also investigates how behavioral and academic factors such as attendance, study hours, parental education, and assignment submission rate influence student success. The results show that ensemble models such as Random Forest and Gradient Boosting achieve higher predictive accuracy than traditional statistical models.
Ultimately, this study demonstrates that integrating ML into educational systems can significantly improve academic planning and decision-making. By identifying learning trends early, institutions can move toward a data-informed educational ecosystem that enhances student engagement and academic performance.
Keywords: Data Science, Machine Learning, Student Performance Prediction, Educational Analytics, Predictive Modeling, Artificial Intelligence, Random Forest.
This research explores how Data Science and Machine Learning (ML) can be applied to predict student academic performance using structured datasets. It highlights algorithms such as Linear Regression, Decision Tree, Random Forest, Support Vector Machine (SVM), and Artificial Neural Network (ANN), demonstrating their potential to analyze educational data and forecast learning outcomes. The study employs Python-based tools such as Scikit-learn, Pandas, and NumPy for model training, testing, and evaluation.
A complete ML pipeline is designed — including data collection, preprocessing, feature selection, model development, and performance evaluation — to predict student grades and categorize learners into performance classes such as High, Medium, and Low. Performance metrics like accuracy, precision, recall, and F1-score are used to evaluate model effectiveness.
The research also investigates how behavioral and academic factors such as attendance, study hours, parental education, and assignment submission rate influence student success. The results show that ensemble models such as Random Forest and Gradient Boosting achieve higher predictive accuracy than traditional statistical models.
Ultimately, this study demonstrates that integrating ML into educational systems can significantly improve academic planning and decision-making. By identifying learning trends early, institutions can move toward a data-informed educational ecosystem that enhances student engagement and academic performance.
Keywords: Data Science, Machine Learning, Student Performance Prediction, Educational Analytics, Predictive Modeling, Artificial Intelligence, Random Forest.
Abstract
Cyber security in 6G network.
Ms. Chetana Kawale*, Dipesh R Patil
DOI: 10.17148/IJIREEICE.2025.131030
Abstract: The Sixth-Generation (6G) network is poised to move beyond mobile internet to enable a seamless fusion of the physical, digital, and human worlds. This paper provides a comprehensive overview of the 6G vision, its key enabling technologies, and the associated challenges. The core objective of 6G is to overcome the limitations of 5G, such as coverage gaps and high infrastructure costs, by targeting unprecedented performance metrics. These metrics include terabit-per-second (Tbps) data rates, sub-millisecond latency (up to microsecond levels), and a massive connection density of up to 10 million devices per square kilometer. Key enabling technologies explored include the utilization of the Terahertz (THz) spectrum, the deep integration of Artificial Intelligence (AI) and Machine Learning (ML) to create self-optimizing networks, and the integration of Sensing and Communication functionalities. Potential applications, such as Holographic Communication, Massive Digital Twins, and Real-time Remote Surgery, are examined. Finally, critical challenges, including spectrum scarcity, energy efficiency, and security threats from quantum computing and AI-driven systems, are discussed to guide future research and development.
Abstract
SMS Spam Classifier
Sura Reddy, Sriram K, Brunda G, Dr. Golda Dilip
DOI: 10.17148/IJIREEICE.2025.131031
Abstract: This project focuses on classifying SMS messages as spam or ham (not spam) using machine learning models. The system collects a labelled dataset of SMS messages, performs text preprocessing (tokenization, stop-word removal, lemmatization), converts text into numerical features using TF-IDF or Word Embeddings, and trains classifiers such as Logistic Regression, Naive Bayes, and SVM. The model achieving the highest accuracy is used for deployment through a Streamlit web app.
Keywords: Machine Learning, NLP, Spam Detection, SMS Classification, TF-IDF, Streamlit
Keywords: Machine Learning, NLP, Spam Detection, SMS Classification, TF-IDF, Streamlit
Abstract
Unified Payments Interface (UPI) System
Prof. Dr. Dinesh D. Puri, Mr. Rohit Narendra Badgujar
DOI: 10.17148/IJIREEICE.2025.131032
Abstract: This project report presents a comprehensive analysis of the UPI system, framed as a large-scale software engineering project, from its conceptualization and planning to its design, implementation, and real-world impact. The system was conceived to address the critical challenges of a cash-dominant economy, aiming to provide a secure, interoperable, and highly convenient mobile-first payment solution. The design is based on a sophisticated four-pillar, API-driven architecture that decouples user-facing applications from the core banking infrastructure, fostering a competitive and innovative ecosystem of Payment Service Providers (PSPs). This model facilitates seamless peer-to-peer (P2P) and person-to-merchant (P2M) transactions using simple identifiers like a Virtual Payment Address (VPA) or QR code, eliminating the need to share sensitive bank account details. Key features include the integration of multiple bank accounts into a single application and mandatory two-factor authentication via a UPI PIN for all transactions. The development and rollout of the system were guided by a hybrid methodology, combining a Waterfall approach for the stable core infrastructure with Agile principles for the rapidly evolving ecosystem of third-party applications. Rigorous integration, performance, and security testing were paramount to ensure the system's reliability and ability to scale to billions of monthly transactions. The results since its 2016 launch have been phenomenal, with UPI now accounting for the vast majority of digital transaction volumes in India.
Abstract
MACHINE LEARNING-BASED NETWORK INTRUSION DETECTION SYSTEM USING THE CSE-CIC-IDS2018 DATASET
Sahithyaa Krishna Kumar, R Sivani, M Jaiaakash, Dr Golda Dilip
DOI: 10.17148/IJIREEICE.2025.131033
Abstract: Network Intrusion Detection Systems (NIDS) are vital defences against evolving and sophisticated cyber threats. Traditional security approaches frequently fail to detect novel, low-volume polymorphic attacks, necessitating the integration of adaptive machine learning (ML) models. This paper presents a high-performance, computationally efficient ML-based NIDS utilizing the contemporary CSE-CIC-IDS2018 dataset. This corpus is preferred over older, synthetic benchmarks (e.g., NSL-KDD) because it provides high-fidelity, B-profile generated benign traffic, ensuring model training accurately reflects real-world network operations. The proposed system employs a Random Forest (RF) classifier, selected for its superior balance of classification accuracy, computational efficiency, and intrinsic feature importance measurement compared to resource-intensive Deep Learning (DL) alternatives.1 The comprehensive methodology includes data cleaning, feature standardization via StandardScaler, and the application of synthetic oversampling techniques (SMOTE) to mitigate the severe class imbalance inherent in network traffic data.3 Experimental results demonstrate that the RF model, optimized via wrapper-based feature selection, achieves a high overall accuracy of 99.9% and robust macro-averaged F1-scores exceeding 96% across seven major attack classes, validating its resilience and practical deploy ability in resource-constrained, large-scale network environments.
Keywords: Network Intrusion Detection, Machine Learning, Random Forest, CSE-CIC-IDS2018, Feature Selection, Class Imbalance, Cybersecurity.
Keywords: Network Intrusion Detection, Machine Learning, Random Forest, CSE-CIC-IDS2018, Feature Selection, Class Imbalance, Cybersecurity.
Abstract
Expert Technical Report: Critical Analysis and Strategic Roadmap for Multi-Agent Q-Learning in Heterogeneous Disaster Response
VIGNESH MURALI, T AAKASH, SHARAN S, Dr GOLDA DILIP
DOI: 10.17148/IJIREEICE.2025.131034
Abstract: Effective disaster management demands rapid coordination between heterogeneous agents tasked with search, rescue, and debris clearance in dynamic environments. Traditional simulation tools often lack flexibility, contextual awareness, and scalability, limiting their use in evaluating multi-agent cooperation under realistic conditions. This paper introduces MAS-SDM (Multi-Agent Simulation Sandbox for Disaster Management), an intelligent, tile-based simulation framework designed to model and analyze autonomous agent behaviour within disaster zones. Built on a 10×10 grid environment created using the Tiled Map Editor, the system simulates a constrained yet richly interactive disaster landscape featuring survivors, debris, safe zones, and obstacles distributed across layered terrain. The sandbox employs four cooperative agents—two specialized in survivor rescue and two in debris removal—each governed by rule-based or reinforcement learning policies that enable dynamic decision-making and task prioritization. Through real-time visualization powered by the Python Pygame engine, MAS-SDM provides an experimental platform for evaluating agent efficiency, coordination strategies, and environment adaptability. Beyond simulating immediate response scenarios, the framework serves as a foundation for developing scalable, data-driven models in multi-agent reinforcement learning (MARL) and disaster logistics optimization. Future work will extend the simulation to larger maps, introduce adaptive communication between agents, and integrate learning modules to autonomously improve cooperative performance in complex, evolving disaster environments.
Keywords: Multi-Agent Systems, Disaster Management, Simulation Sandbox, Tile-based Environment, Reinforcement Learning, Cooperative Agents, Search and Rescue, Debris Clearance, Tiled Map Editor, Pygame
Keywords: Multi-Agent Systems, Disaster Management, Simulation Sandbox, Tile-based Environment, Reinforcement Learning, Cooperative Agents, Search and Rescue, Debris Clearance, Tiled Map Editor, Pygame
Abstract
Full-Stack Implementation and Evaluation of Abstractive Text Summarization using a Transformer-Based BART Model
MARK OWEN A, PAARIVALAVAN S, SANJAY C, Dr GOLDA DILIP
DOI: 10.17148/IJIREEICE.2025.131035
Abstract: The exponential growth of online textual data has created a critical need for efficient information processing, making automated text summarization an indispensable tool for mitigating information overload. While traditional methods have struggled with fluency and coherence, the advent of Transformer-based models has defined a new state- of-the-art. This paper introduces a robust, end-to-end framework for abstractive text summarization leveraging a pre- trained BART (Bidirectional and Auto-Regressive Transformers) model. The system utilizes the facebook/bart-large-cnn model, a specific variant fine-tuned on the CNN/Daily Mail news dataset, which employs a bidirectional encoder for comprehension and an auto-regressive decoder to generate novel text. This AI core is deployed within a modern, scalable web application, served via a high-performance FastAPI backend API and consumed by an interactive React user interface. This paper details the full-stack architecture, from the model-loading strategy at server startup to the asynchronous API request-response workflow. The model's performance is quantitatively evaluated using the standard ROUGE metrics, demonstrating strong results with a mean ROUGE-1 F1 score of 50.61% and a ROUGE-L F1 score of 42.99%. We provide a detailed analysis of these metrics, including precision/recall trade-offs and score distributions, confirming the model's high abstractive capability. This research serves as a comprehensive blueprint for the practical implementation and evaluation of a state-of-the-art Transformer approach for real-world summarization applications.
Keywords: Abstractive Text Summarization, BART, Transformer, Natural Language Processing (NLP), ROUGE, FastAPI, React, Full-Stack Application, CNN/Daily Mail.
Keywords: Abstractive Text Summarization, BART, Transformer, Natural Language Processing (NLP), ROUGE, FastAPI, React, Full-Stack Application, CNN/Daily Mail.
Abstract
COMPUTER VISION: TEXT EXTRACTION FROM AN IMAGE
Yokesh Anandan, Christon Davis C, Dr. Golda Dilip
DOI: 10.17148/IJIREEICE.2025.131036
Abstract: With the proliferation of visual data, the automatic extraction of text from images using Optical Character Recognition (OCR) has become a crucial application in computer vision. This tutorial demonstrates a rapid and effective method for text detection and extraction using Python, the EasyOCR library, and OpenCV. The core pipeline involves reading an image, creating an instance of the text detector, processing the text, and visualizing the results by drawing bounding boxes around the detected text. The resulting algorithm is a quick and ideal project for beginners in computer vision, completing the process in only a few minutes.
Keywords: Computer Vision, Optical Character Recognition, Text Detection, Python, EasyOCR, OpenCV, Bounding Box.
Keywords: Computer Vision, Optical Character Recognition, Text Detection, Python, EasyOCR, OpenCV, Bounding Box.
Abstract
MUSIC GENRE CLASSIFICATION SYSTEM USING NATURAL LANGUAGE PROCESSING
PRANAV H, SARVESH S, HARI SRINIVAS, Dr. Golda Dilip
DOI: 10.17148/IJIREEICE.2025.131037
Abstract: In this paper, we present a Music Genre Classification System that utilizes Natural Language Processing (NLP) and Machine Learning techniques to predict the genre of a song solely based on its lyrical content. Unlike traditional audio-based classification methods, this approach focuses on the textual features of lyrics, enabling faster and more resource-efficient analysis. The system begins with data acquisition from publicly available song lyrics datasets, followed by rigorous text preprocessing involving tokenization, Stopword removal, and lemmatization to standardize input data. Feature extraction is performed using the Term Frequency–Inverse Document Frequency (TF-IDF) technique to represent textual information numerically, preserving the contextual importance of words.
The processed data is then used to train a supervised machine learning model, specifically Logistic Regression, which learns distinctive linguistic and stylistic patterns associated with different genres such as Pop, Rock, Hip-Hop, and Country. Model evaluation was carried out using metrics like accuracy, precision, recall, and F1-score, achieving an overall accuracy of approximately 85%. A user-friendly web interface was developed using Streamlit to allow real-time lyric input and instant genre prediction.
The proposed system demonstrates that lyrics carry significant semantic and emotional information that can be leveraged to classify music genres effectively. This work contributes to the growing field of computational music analysis and can be further extended to enhance music recommendation engines, automated playlist generation, and text-based sentiment- driven music analysis.
The processed data is then used to train a supervised machine learning model, specifically Logistic Regression, which learns distinctive linguistic and stylistic patterns associated with different genres such as Pop, Rock, Hip-Hop, and Country. Model evaluation was carried out using metrics like accuracy, precision, recall, and F1-score, achieving an overall accuracy of approximately 85%. A user-friendly web interface was developed using Streamlit to allow real-time lyric input and instant genre prediction.
The proposed system demonstrates that lyrics carry significant semantic and emotional information that can be leveraged to classify music genres effectively. This work contributes to the growing field of computational music analysis and can be further extended to enhance music recommendation engines, automated playlist generation, and text-based sentiment- driven music analysis.
Abstract
Machine Learning-Based Plagiarism Detector System
Rakshitha S N, Shreya Sathapathi, Yashvitha J, Dr. Golda Dilip
DOI: 10.17148/IJIREEICE.2025.131038
Abstract: Robust tools for plagiarism checking are essential for maintaining integrity in both academic and professional environments. Existing detection strategies, which are typically built on lexical comparison, struggle to correctly flag sophisticated, machine-aided rephrasing. This challenge requires a necessary pivot toward adaptable Machine Learning (ML) platforms capable of comprehending the underlying meaning of text. This research introduces a highly efficient, two-phase ML framework specifically engineered to accurately identify text that has been heavily paraphrased. The initial phase of this architecture employs a SentenceTransformer model (all-MiniLM-L6-v2) to generate dense vector embeddings for documents under suspicion and for the reference library. These embeddings are stored and searched using FAISS (Facebook AI Similarity Search), enabling fast, large-scale retrieval of potential source candidates. The second phase uses a Longformer-based sequence classifier to perform an in-depth, pairwise contextual analysis between the flagged text and the retrieved candidates before delivering a final verdict. This classifier model was chosen because it effectively bypasses the sequence-length constraints of previous transformer models, enabling analysis of long-form content. The final system, named "CopyShield," is deployed with an accessible user interface using the Gradio framework. Validation using the challenging jpwahle/machine-paraphrase-dataset demonstrated a strong F1-score in the 0.89–0.92 range, confirming its ability to counter contemporary obfuscation methods.
Keywords: NLP, ML, Semantic Analysis, Transformer Models, Deep Learning, Longformer Architecture, Plagiarism Checkers, Gradio, FAISS.
Keywords: NLP, ML, Semantic Analysis, Transformer Models, Deep Learning, Longformer Architecture, Plagiarism Checkers, Gradio, FAISS.
Abstract
Enhancing Prediction and Explainability with Machine Learning Using SHAP on OASIS MRI Data Compared to Traditional Diagnosis Methods
Mrinmayi Verma, Neelam Sanjeev Kumar
DOI: 10.17148/IJIREEICE.2025.131039
Abstract: Alzheimer’s disease (AD) has emerged as a significant health challenge globally, with projections reaching over 150 million affected individuals by 2050. Early diagnosis remains pivotal in managing disease progression and improving patient quality of life. Traditional diagnostic techniques rely heavily on neuropsychological assessments and qualitative MRI analysis, which suffer from subjective biases and inter-observer variability, often delaying diagnosis or leading to inaccuracies (Marcus et al., 2007; Marcus et al., 2010).
Recent breakthroughs in machine learning (ML), especially ensemble models combined with explainability techniques like SHAP (SHapley Additive exPlanations), have penned a new era in medical diagnostics where models can be both accurate and transparent (Lundberg & Lee, 2017). Our approach leverages Random Forest classifiers trained on the OASIS dataset—comprising heterogenous, multimodal data including MRIs, clinical scores, and demographics. The model’s decision process is elucidated through SHAP, allowing clinicians to understand the relative importance of features such as regional brain atrophy, age, and cognitive scores, thus aligning model outputs with biological plausibility and increasing clinical trust.
Furthermore, spatial localization through Grad-CAM overlays provides anatomical context to model decisions, highlighting brain regions like hippocampus and temporal lobes that are traditionally associated with AD (Selvaraju et al., 2017). This combined approach exemplifies a transparent, high-performing framework compatible with clinical workflows, offering a benchmark for future multi-modal, explainable AI models for neurodegenerative diseases, and emphasizes the road toward trustworthy AI-driven diagnostics that reconcile accuracy with interpretability (Mahavar et al., 2025).
Keywords: Alzheimer’s disease, MRI, Random Forest, SHAP, explainable AI, OASIS, ensemble learning, Grad-CAM.
Recent breakthroughs in machine learning (ML), especially ensemble models combined with explainability techniques like SHAP (SHapley Additive exPlanations), have penned a new era in medical diagnostics where models can be both accurate and transparent (Lundberg & Lee, 2017). Our approach leverages Random Forest classifiers trained on the OASIS dataset—comprising heterogenous, multimodal data including MRIs, clinical scores, and demographics. The model’s decision process is elucidated through SHAP, allowing clinicians to understand the relative importance of features such as regional brain atrophy, age, and cognitive scores, thus aligning model outputs with biological plausibility and increasing clinical trust.
Furthermore, spatial localization through Grad-CAM overlays provides anatomical context to model decisions, highlighting brain regions like hippocampus and temporal lobes that are traditionally associated with AD (Selvaraju et al., 2017). This combined approach exemplifies a transparent, high-performing framework compatible with clinical workflows, offering a benchmark for future multi-modal, explainable AI models for neurodegenerative diseases, and emphasizes the road toward trustworthy AI-driven diagnostics that reconcile accuracy with interpretability (Mahavar et al., 2025).
Keywords: Alzheimer’s disease, MRI, Random Forest, SHAP, explainable AI, OASIS, ensemble learning, Grad-CAM.
Abstract
Advanced Machine Learning for Real-Time Driver Distraction Analysis with Visual Inputs
Ramisetty T M Surya, Sai G, Allam Reddy Charan, Sriram Sanjay S, Neelam Sanjeev Kumar
DOI: 10.17148/IJIREEICE.2025.131040
Abstract: Driving can be risky when a driver’s mind wanders, even if their eyes are on the road. This “look but don’t see” problem, called cognitive distraction, is a major cause of car crashes. As self-driving cars become more common, humans will still need to stay alert to take control in emergencies for years to come. To tackle this, we’ve developed a new model called Self-DSNet to detect when drivers are distracted. Self-DSNet uses a special kind of neural network to spot complex patterns in data. When tested with just camera footage, it was 94.23% accurate at catching distractions. Adding data like heart rate, breathing rate, and how the driver steers the car boosted accuracy to 95.13%. The model relies on using tools like Random Forest, Decision Trees, and Support Vector Machines to make its predictions. We also found that focusing on just a few key signs—like changes in a driver’s pupil size or eye movements—still gave solid results, with 90% accuracy across different types of roads. The study also showed that the type of road can affect how distracted a driver gets. These findings could help build better systems to keep drivers focused. In the future, researchers plan to test this model in real-time driving situations and add more data sources to make it even more reliable across all kinds of roads and scenarios.
Keywords: Driver distraction, Human driving supervision, Vehicle sensors, Self-DSNet model, Self-Organizing Neural Network (Self-ONN), Driver distraction monitoring systems.
Keywords: Driver distraction, Human driving supervision, Vehicle sensors, Self-DSNet model, Self-Organizing Neural Network (Self-ONN), Driver distraction monitoring systems.
Abstract
Hybrid Deep Learning Methodology For Disease Classification In Medical Imaging
Ms. Charulatha RT, M. Adethya, Mani Aadithyaa A.A, Jeevan P, Anto Nawin
DOI: 10.17148/IJIREEICE.2025.131041
Abstract: Medical imaging forms a cornerstone of modern diagnostic healthcare; still manual image interpretation is intensive and susceptible to discrepancies among different observers. This work introduces a hybrid deep learning methodology that combines Convolutional Neural Networks (CNNs) with classical image processing techniques to classify chest X-rays into diseased and normal categories. Using transfer learning with ResNet50 and feature fusion involving edge, corner, and texture descriptors, the suggested architectural framework exhibits enhanced efficacy in the detection of pneumonia, tuberculosis, and COVID-19. The dataset used is NIH Chest X-ray14 and supplementary datasets show enhanced accuracy, recall and area under the curve. Furthermore, explainability tools such as Grad-CAM overlays with heat maps provide interpretability and clinical confidence, addressing a major gap in AI-assisted diagnostics.
Keywords: Deep Learning, Medical Imaging, CNN, Hybrid Model, Chest X-ray, Explainable AI, Transfer Learning, Grad- CAM, Multi-label Classification.
Keywords: Deep Learning, Medical Imaging, CNN, Hybrid Model, Chest X-ray, Explainable AI, Transfer Learning, Grad- CAM, Multi-label Classification.
Abstract
PHISHING DETECTION USING CNN AND BiLSTM
Princeton Vishal J, Srinath S, Adithya S, Saran P, Yogesh C, Neelam Sanjeev Kumar
DOI: 10.17148/IJIREEICE.2025.131042
Abstract: Phishing continues to be one of the most deceptive and persistent cyber threats in today’s interconnected digital landscape, targeting unsuspecting users through fraudulent websites and emails designed to steal sensitive information. This study presents a robust hybrid deep learning model that integrates Convolutional Neural Networks (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) architectures to enhance phishing detection accuracy and resilience. The CNN layers effectively capture spatial and lexical patterns from URLs and email content, while the BiLSTM layers analyze sequential dependencies and contextual relationships within textual data. Together, these components enable the model to learn both structural and semantic cues associated with phishing behavior.Experimental evaluations conducted on benchmark phishing datasets demonstrated that the proposed hybrid CNN-BiLSTM model achieved an overall detection accuracy exceeding 95%, outperforming traditional machine learning algorithms such as SVM and Random Forest. The system also showed superior precision and recall, reducing false positives and improving interpretability through an integrated attention mechanism. This research contributes to the advancement of cybersecurity by proposing an adaptive, data-driven defense framework capable of evolving alongside emerging phishing strategies and offering practical potential for real-time threat mitigation.Keywords: Phishing Detection, CNN, BiLSTM, Deep Learning, Cybersecurity.
Abstract
MACHINE LEARNING PREDICTION FOR CROP SELECTION USING A RANDOM FOREST CLASSIFIER
Sreenidhi T, Mithra B V, Lathika M, Kaviya S, Ahana Dinesh, Neelam Sanjeev Kumar
DOI: 10.17148/IJIREEICE.2025.131043
Abstract: This work proposes a machine learning crop recommendation system that supports enhanced agricultural decision making. The model employs crucial factors such as soil nutrients (nitrogen, phosphorus, potassium), pH, TMP, HUM, and RF in order to predict optimal crops to plant. We compared five classifiers: Logistic Regression, Support Vector Classification (SVC), Multilayer Perceptron (MLP), Random Forest and Decision Tree. Among these models, RF performed the best with the highest accuracy of 99.27%, demonstrating good performance on challenging agricultural data. The model was released as an interactive web app developed in Streamlit, which generates a real-time forecast for farmers to choose a crop. This study has shown the superiority of ensemble and nonlinear machine learning models over linear type in agriculture. It offers a potential scaling tool toward enhancing crop yield and sustainability.
Keywords: Machine Learning, Crop Selection System, Smart Agriculture, Random Forest Classifier, Performance Analysis, Scalability, Streamlit Web Application, Prediction Accuracy.
Keywords: Machine Learning, Crop Selection System, Smart Agriculture, Random Forest Classifier, Performance Analysis, Scalability, Streamlit Web Application, Prediction Accuracy.
Abstract
Hybrid PSO-CNN-LSTM Framework for Intelligent DDoS Detection
Dinesh P, Eedpuganti Yagna Sai Harshith, Athithya S A, Raj Pranav Raghavan,G Mohammed Azam, Neelam Sanjeev Kumar
DOI: 10.17148/IJIREEICE.2025.131044
Abstract: In modern network environments, Distributed Denial-of-Service (DDoS) attacks represent a critical security threat, often rendering services unavailable to legitimate users. Conventional intrusion detection techniques fail to adapt to evolving attack strategies and high-dimensional network traffic. This paper proposes a PSO-optimized CNN-LSTM hybrid deep learning model for accurate DDoS detection and classification. The CNN component extracts spatial features from traffic flows, while the LSTM component captures temporal dependencies in sequential data. Particle Swarm Optimization (PSO) is employed to optimize hyperparameters, enabling faster convergence and improved performance. The model is trained and evaluated on imbalanced real-world network datasets. Experimental results indicate significant improvements, achieving over 98% accuracy and enhanced recall and F1-score compared to traditional models. The proposed method demonstrates strong potential for deployment in real-time cybersecurity systems, offering robustness, automation, and adaptability in detecting DDoS attacks.
Keywords: DDoS Detection, CNN-LSTM, Deep Learning, Particle Swarm Optimization (PSO), Cybersecurity, Network Intrusion Detection, Real-Time Security, Imbalanced Dataset, Hyperparameter Optimization, Spatial-Temporal Learning
Keywords: DDoS Detection, CNN-LSTM, Deep Learning, Particle Swarm Optimization (PSO), Cybersecurity, Network Intrusion Detection, Real-Time Security, Imbalanced Dataset, Hyperparameter Optimization, Spatial-Temporal Learning
Abstract
Adaptive Control Strategies for Human-Robot Interaction in Industrial Setting
Nikunj Kaslikar, Aarav Singh, Jeevasree S, Charuhazan B, Neelam Sanjeev Kumar
DOI: 10.17148/IJIREEICE.2025.131045
Abstract: Human–Robot Interaction (HRI) represents a crucial frontier in modern robotics, enabling robots to collaborate intelligently and safely with humans across industrial, medical, and service environments. However, dynamic human behaviour, unpredictable environmental conditions, and task variability pose significant challenges to achieving seamless interaction. This study introduces a novel Adaptive Control Strategy (ACS) framework designed to enhance the responsiveness, safety, and efficiency of HRI systems. The proposed approach integrates reinforcement learning, fuzzy logic control, and model predictive control (MPC) to enable robots to dynamically adjust their motion, force, and communication behaviour based on continuous feedback from human partners and environmental sensors.
The adaptive framework allows the robot to learn and anticipate human intentions in real time through multimodal sensory fusion, combining vision, force, and voice data streams. By employing online learning and parameter tuning, the control system ensures smooth trajectory tracking, minimizes physical and cognitive workload on humans, and prevents unsafe interactions. Experimental evaluations were conducted using both simulated and real-world HRI scenarios involving cooperative manipulation and shared workspace tasks. The results demonstrate that the proposed adaptive control model achieves significant performance improvements, including faster response adaptation, reduced interaction delays, and enhanced stability compared to conventional fixed-gain and non-adaptive controllers.
Furthermore, statistical analysis indicates that the system achieves interaction accuracy above 96%, maintaining robust performance under uncertain human motions and external disturbances. The adaptive nature of the controller allows it to generalize across diverse human behaviour without explicit reprogramming, thereby improving scalability and usability. This research contributes to the advancement of intelligent robotic systems by presenting a human-centered, data-driven control architecture that evolves continuously, fostering safe, natural, and efficient collaboration between humans and robots in real-world settings.
Keywords: Adaptive Control, Human–Robot Interaction, Reinforcement Learning, Fuzzy Logic, Model Predictive Control, Intelligent Robotics.
The adaptive framework allows the robot to learn and anticipate human intentions in real time through multimodal sensory fusion, combining vision, force, and voice data streams. By employing online learning and parameter tuning, the control system ensures smooth trajectory tracking, minimizes physical and cognitive workload on humans, and prevents unsafe interactions. Experimental evaluations were conducted using both simulated and real-world HRI scenarios involving cooperative manipulation and shared workspace tasks. The results demonstrate that the proposed adaptive control model achieves significant performance improvements, including faster response adaptation, reduced interaction delays, and enhanced stability compared to conventional fixed-gain and non-adaptive controllers.
Furthermore, statistical analysis indicates that the system achieves interaction accuracy above 96%, maintaining robust performance under uncertain human motions and external disturbances. The adaptive nature of the controller allows it to generalize across diverse human behaviour without explicit reprogramming, thereby improving scalability and usability. This research contributes to the advancement of intelligent robotic systems by presenting a human-centered, data-driven control architecture that evolves continuously, fostering safe, natural, and efficient collaboration between humans and robots in real-world settings.
Keywords: Adaptive Control, Human–Robot Interaction, Reinforcement Learning, Fuzzy Logic, Model Predictive Control, Intelligent Robotics.