| 1 |
Author(s):
Emiliano C. De Catalina.
Country:
Philippines
Research Area:
Electrical Engineering
Page No:
1-11 |
Electrical Method in Automated On-Off and Speed-Selection Switching of 220V AC Electric Fan
Abstract
This paper is concerned with the 220VAC electric fan. Most electric fans sold in the market today are still manually operated, except for a few automated ones. Mostly, the ON-OFF and SPEED-SELECTION switching is still manually done by the user. The purpose of this paper is to present an innovation on the said manually-operated ON-OFF and SPEED-SELECTION switching of the common household 220VAC electric fan. Here, the switching is to be automated by means of ambient temperature; but no microcontroller, with its program, no internet would be used. Generally, this innovation is an automated ON-OFF and SPEED-SELECTION switching using electrical method. As temperature rises, the electric fan itself automatically switches ON and selects to the higher SPEED. As temperature lowers, the electric fan also automatically selects to the lower SPEED, corresponding to a pre-set temperature, and, switches OFF as temperature lowers even more. This study utilizes the applied technological research design. It applies scientific knowledge to solve a problem or to develop a new method. The findings of this study show that the innovation is feasible, technically doable, and implementable. In conclusion, the innovation is manufacturable, integrable, and adoptable in the current designs of the 220VAC electric fans commonly sold in the market today.
| 2 |
Author(s):
Harshdeep Mishra, Gagan Sharma, Nidhi Sharma, Ashutosh Tripathi.
Country:
India
Research Area:
Computer Science
Page No:
12-23 |
AI-Based Predictive Waste Reduction System in Supermarkets: A Dashboard-Driven Approach
Abstract
Food waste represents one of the significant global sustainability challenges, with supermarkets being substantial contributors due to inadequacies in demand forecasting and inventory management. This paper presents an AI-Based predictive waste reduction system designed for supermarket environments. The system integrates machine learning-driven risk scoring, real-time inventory monitoring, and an interactive analytics dashboard to identify high-risk products and deliver actionable recommendations such as dynamic discounting, stock redistribution, and donation management. Built on a modular React.js/TypeScript architecture with Tailwind CSS styling, the platform offers multi-store support, configurable alert thresholds, and a responsive user interface. Empirical testing demonstrates the system’s capacity to reduce food waste by up to 47% and generate daily savings of approximately $2,847/store through proactive interventions. The paper details the system architecture, feature design, database schema, and evaluation results. It also looks ahead to real world use, highlighting how the system could relate to IoT sensor and integrated with cloud-based machine learning to make it practical, scalable and ready for use.
| 3 |
Author(s):
Sagar Pathak, Bidhya Shrestha.
Country:
United States
Research Area:
Computer Science
Page No:
24-39 |
Enhanced Deep Q-Learning for 2D Self-Driving Cars: Implementation and Evaluation on a Custom Track Environment
Abstract
This research presents the implementation of a Deep Q-Learning Network (DQN) for a self-driving car on a 2-dimensional (2D) track, aiming to enhance the performance of the DQN network. It covers the development of custom driving environment with pygame on the track around the University of Memphis map and design and implementation of the DQN model. The algorithm utilizes data from 7 sensors collected by sensors installed in the car, based on the distance between the car and the track. These sensors are positioned in front of the vehicle, spaced 20 degrees apart, enabling them to sense a wide area ahead. We successfully implemented DQN and also modified the DQN with priority-based action selection mechanism and referred to it as modified DQN. The model is trained on 1000 episodes and the average reward received by agent is found to be around 40, which is around 60% higher than the original DQN and around 50% higher than the vanilla neural network.
| 4 |
Author(s):
K. D. Jagtap, B. D. Karande, S. V. Badgire.
Country:
India
Research Area:
Mathematics
Page No:
40-57 |
On Existence and Locally Attractivity Results for Fractional Order Nonlinear Random Integral Equation
Abstract
In this paper, we investigate the existence and qualitative behaviour of solutions for a class of nonlinear random integral equation of fractional order in R_+=[0,∞). The analysis is carried out within the framework of Banach algebra, employing a hybrid fixed-point theorem as the principal tool. The problem is considered under the assumptions of Lipschitz continuity and Caratheodory conditions, which ensures the measurability and continuity properties required for the existence of random solutions. In addition to proving the existence of such solutions, we establish their local attractivity, thereby demonstrating the stability of the system in a probabilistic sense. Also we have proved existence of extremal solutions. The theoretical results presented in this work contribute to the growing field of fractional calculus and stochastic analysis by providing rigorous framework for studying fractional random integral equations. To illustrate the applicability of the main results, we provide a concrete example that verifies the theoretical findings and highlights the practical relevant of the proposed approach.
| 5 |
Author(s):
Kufre A. Mkpedem, Mary O. Durojaye.
Country:
United Kingdom
Research Area:
Mathematics
Page No:
58-71 |
Polynomial Approximation Analysis of Transient Mixed Convection Flow in A Vertical Micro-Annulus with Viscous Dissipation
Abstract
This paper presents an analytical assessment of transient mixed convection flow and heat transfer in a vertical micro-annulus, taking into account internal heat generation, viscous dissipation, and temperature-dependent viscosity. The Boussinesq approximation is used to develop the governing momentum and energy equations, which are then translated into dimensionless form. A polynomial approximation method is used to generate closed-form solutions, which are then evaluated via symbolic computation. The effect of critical dimensionless factors, such as the Reynolds number, Peclet number, Eckert number, and heat generation coefficients, on fluid velocity and temperature distributions is investigated. The findings show that internal heat generation greatly increases fluid temperature while decreasing flow velocity. Increasing Eckert and Peclet numbers reduces temperature profiles, while increasing Reynolds numbers decreases fluid velocity. The research also reveals a linked interaction between heat and flow fields, which is controlled by viscosity variations and energy dissipation effects. The findings have important implications for thermal management and flow control in micro-scale systems including microchannels, heat exchangers, and cooling devices, where accurate prediction of linked heat and fluid flow is critical.
| 6 |
Author(s):
Shatakshi Singh, Satwick Pandey, Smriti Jaiswal.
Country:
India
Research Area:
Computer Science
Page No:
72-84 |
AI-Enabled Smart Energy Optimization System for Consumption Forecasting and Cost Reduction
Abstract
Managing household electricity consumption in India is challenging due to complex slab-based tariff structures and limited consumer awareness of appliance-level energy usage. This paper presents a Next Generation Smart Energy Optimizer, an AI-driven web-based system designed to predict electricity consumption and optimize household energy costs. The proposed system integrates machine learning, a slab-aware billing engine, and large language model (LLM)-based recommendations to provide comprehensive decision support. A Random Forest regression model, trained on real-world household data, predicts appliance-wise monthly energy consumption with high accuracy. The billing engine simulates real Distribution Company (DISCOM) tariff structures, generating detailed cost breakdowns including energy charges, fixed charges, and applicable duties. Additionally, an LLM-based advisory module produces personalized and context-aware recommendations to help users reduce consumption and expenses. The system is implemented using a Python–Flask backend, PostgreSQL database, and a browser-based frontend, ensuring scalability and accessibility. Experimental evaluation shows strong performance, achieving a mean absolute percentage error (MAPE) of 6.8% compared to actual electricity bills. The proposed framework bridges the gap between consumption prediction and actionable insights, offering a practical solution for intelligent household energy management.
| 7 |
Author(s):
Maribel M. Estioco, Wences Love Ylanan, Jezreel Roy Mamhot, Jeza Senining, Erica Pearl Faith Sarueda.
Country:
Philippines
Research Area:
Information Technology
Page No:
85-98 |
Development and Performance Evaluation of a Web-Based Facial Recognition Attendance System for Higher Education Institutions
Abstract
The web-based facial recognition attendance system, designed and evaluated in this project, provided an efficient, accurate, and secure means of taking attendance at J.H. Cerilles State College. Face-to-face identification can eliminate errors in manual and pen-and-paper recording, such as late manual logs and logbook signing for absent students. This system utilizes the 4D development model: Discover, Design, Develop, and Deploy for automating this traditional process. The web system is built with HTML, CSS, JavaScript, PHP, MySQL, and face-api.js for browser-based real-time face detection and recognition. Security for user login, face enrollment, automatic attendance recording, role-based management, and record export are among the main features of this system. In terms of system evaluation, the tests undertaken included alpha, beta, load, stress, compatibility, usability, and functionality testing. Analysis of these tests showed high user acceptance. The total weighted mean across these tests was 4.73, corresponding to Strongly Agree. The system demonstrated consistent response time and maintained accurate recognition performance with 30 users concurrently operating. Stress testing successfully identified 249/265 users, achieving 93.96% recognition accuracy during peak usage. Compatibility testing revealed that the system functions correctly in current web browsers like Google Chrome and Brave and includes proper privacy measures. The result shows that the system reduced the workload on admin staff, reduced attendance errors, and increased the transparency of the attendance system. We conclude that the web-based facial recognition attendance system is suitable and flexible to apply in a higher education system.