Hydrogen Yield Efficiency Based on Current Density in AWE
Hydrogen production through alkaline water electrolysis (AWE) remains one of the most reliable and economically feasible pathways for generating clean hydrogen. However, the efficiency of AWE is strongly influenced by the operating current density, particularly at higher loads, where bubble accumulation, increased overpotential, and mass-transport limitations reduce the practical hydrogen yield. This study examines the effect of varying current density on hydrogen yield efficiency by comparing the experimentally collected hydrogen volume with theoretical values derived from Faraday’s law. Electrolysis was performed using stainless steel electrodes in a 0.50 M NaOH electrolyte over current inputs ranging from 0.10 A to 0.50 A. For each current setting, the corresponding hydrogen volume was measured via the water displacement method, converted to moles using the ideal gas law, and evaluated against the predicted stoichiometric output. The results show a near-linear increase in hydrogen production at lower current densities but a noticeable deviation from ideal Faradaic behaviour at higher currents. Faradaic efficiency decreased from approximately 94% at 0.10 A to around 85% at 0.50 A, confirming that bubble blockage, resistive heating, and kinetic limitations become more pronounced as current density increases. The study provides a clear, empirical relationship between current density and hydrogen yield efficiency in a simple AWE system, offering useful insights for small-scale electrolysis applications and highlighting the practical limitations encountered when transitioning to higher operational loads. Beyond quantifying efficiency trends, this study also demonstrates the importance of understanding electrochemical behaviour when scaling up hydrogen production systems. Since many educational and laboratory AWE setups operate without advanced engineering features—such as forced electrolyte circulation, porous electrodes, or catalytic coatings—the findings provide a realistic baseline for performance expectations in simple electrolyzers. The observations reinforce that while increasing current density boosts hydrogen output, it simultaneously introduces non-idealities that lower conversion efficiency. These insights can support future optimisations in electrode design, electrolyte composition, and cell configuration for improved hydrogen yield in low-cost AWE systems. Overall, the study highlights the value of Faradaic efficiency as a diagnostic tool for evaluating real-world electrolyzer performance. By directly comparing theoretical and experimental hydrogen yields, the method used here provides a simple yet powerful way to identify operational losses without requiring advanced instrumentation. This approach can be applied in future work to assess the influence of factors such as electrode spacing, electrode surface treatment, electrolyte concentration, and temperature on hydrogen output. The findings, therefore, not only document the behaviour of AWE under varying current densities but also establish a practical framework for improving system efficiency in academic, laboratory, and introductory research settings.
Published by: Harmaya Thukral
Author: Harmaya Thukral
Paper ID: V11I6-1278
Paper Status: published
Published: December 16, 2025
Geospatial Threat Assessment: Safety Analytics using RNN, XGBoost and Isolation Forest
Personal safety applications today mostly respond after an incident occurs, which limits their ability to prevent harm. In this work, we develop a proactive safety-risk prediction system that estimates how dangerous a location may become in the near future. The system combines sequential deep-learning models with boosted decision-tree techniques to understand how local crime risk evolves over time and space. Historical crime records, temporal patterns, nearby points of interest, and environmental context are merged into structured spatio-temporal data grids. The proposed approach uses an LSTM network to learn short-term temporal changes in risk at the grid-cell level, while an XGBoost model evaluates spatial and contextual factors to produce interpretable risk scores. An Isolation Forest module is used alongside these models to detect sudden, unusual conditions that may indicate unsafe situations. The outputs of all three models are merged into a unified risk score that updates continuously and highlights emerging danger zones. When the score crosses certain thresholds, the system can issue early warnings, suggest safer travel routes, or escalate alerts if needed. The system is evaluated on real crime datasets using spatio-temporal cross-validation, and performance is measured using metrics suited for imbalanced data such as AUC, Precision@K, and F1-score. Results demonstrate that the system can provide meaningful early-risk signals while maintaining transparency and privacy-aware processing.
Published by: Mukul Malviya, Mohak Pandagre, Kushagra Singh Chouhan, Mayank Dhakar
Author: Mukul Malviya
Paper ID: V11I6-1228
Paper Status: published
Published: December 16, 2025
Greenscan: An AI-Powered, Cross-Platform System for Instant Plant Identification and Care Guidance
This paper presents GreenScan, an intelligent and interactive web platform developed to enable fast, accurate, and user-friendly plant species recognition through uploaded images. Addressing the persistent challenges of manual plant identification, such as inefficiency, limited accessibility, and a lack of centralized information, GreenScan leverages the power of Artificial Intelligence (AI) and Deep Learning to deliver real-time classification of more than 100 distinct plant species. The system employs a Convolutional Neural Network (CNN) model trained on a large and diverse dataset of plant images to ensure high recognition accuracy, even under varying lighting and background conditions. The platform integrates a responsive and intuitive web interface, allowing users to seamlessly upload images, view classification results, and explore detailed plant profiles. Each identified species is linked to a comprehensive backend database containing essential details such as taxonomy, physical characteristics, ideal growing conditions, and care guidelines. Furthermore, GreenScan provides external purchase links and educational resources, making it an invaluable tool for students, researchers, horticulturists, and nature enthusiasts. A key feature of GreenScan is its feedback-driven learning mechanism, which enables continuous model retraining based on user input to progressively enhance prediction precision over time. The platform’s implementation achieved high confidence scores, including a 91% accuracy rate for identifying species such as the Snake Plant. Beyond its technical merits, GreenScan contributes significantly to promoting environmental education, sustainable living, and ecological awareness by bridging the gap between modern technology and biodiversity knowledge. This work demonstrates the potential of AI-powered solutions to transform traditional plant identification into a more engaging, efficient, and educational digital experience.
Published by: Vaishnavi Duratkar, Twinkal Sapate, Snehal Ninawe, Sanket Barapatre, Ashwary Dhakate, Sharwari Mohadikar, Prajakta Singham, Mamta Balbudhe
Author: Vaishnavi Duratkar
Paper ID: V11I6-1296
Paper Status: published
Published: December 13, 2025
AI-Driven Smart Air Quality Monitoring and Predictive Pollution Control System Using IoT and Edge Computing
Air pollution has become a major environmental threat, yet traditional monitoring systems rely on expensive fixed stations with limited coverage and delayed reporting. This project introduces an AI-driven Smart Air Quality Monitoring and Predictive Pollution Control System that integrates IoT sensors, edge computing, machine learning, and cloud analytics for real-time, scalable monitoring. A network of low-cost sensors measures pollutants such as PM2.5, PM10, CO₂, CO, and NO₂, sending data to an edge device (ESP32/Raspberry Pi) for cleaning, filtering, and anomaly detection. Edge processing minimizes latency, saves bandwidth, and enables rapid local decision-making. Cleaned data is then uploaded to the cloud, where models like Random Forest, XGBoost, and LSTM generate short- and long-term pollution forecasts. An interactive dashboard visualizes real-time AQI, spatial patterns, and predictive insights to support timely interventions. Overall, this cost-effective system demonstrates key CS engineering skills and offers a practical framework for smarter, healthier, and more resilient cities.
Published by: Daksh Jain
Author: Daksh Jain
Paper ID: V11I6-1280
Paper Status: published
Published: December 13, 2025
Parkinson’s Disease – History, detection, and cure
This paper talks about the neurodegenerative disease commonly known as Parkinson's disease, about the history of this disease and its causes, various subtypes, as well as how the diagnosis of this disease takes place. It has mainly 3 causes those being environment, genetics, or interactions. It usually happens to people above the age of 60, but there are younger cases as well. This disease causes loss of dopaminergic neurons in the brain; these neurons help in motor activities of the body, thus their loss causes loss of motor activities of the body. The dopamine-producing neurons Substantia Nigra are directly affected. The neurons degenerate due to the accumulation of Alpha’s nuclein in the brain. The paper discusses how motor, non-motor and psychological aspects should be taken into account during the identification of this disease.
Published by: Prisha Teotia
Author: Prisha Teotia
Paper ID: V11I6-1272
Paper Status: published
Published: December 13, 2025
AI Bias in Data Training
This research paper talks about AI bias in data training and how it creates age, gender and cultural discrimination. This paper also talks about how spreading awareness about AI bias can help mitigate the issue. It examines how biased training data distorts decision-making in various fields like hiring, healthcare and law enforcement. This paper shows us the need for transparency, accountability and awareness in AI systems and how mitigating data bias is essential for creating an AI system that is fair, responsible, and that can be held accountable in case of any biased decisions and output.
Published by: Sarjas Gauhar Singh
Author: Sarjas Gauhar Singh
Paper ID: V11I6-1287
Paper Status: published
Published: December 12, 2025
