-
Pictorial Overview.
Inspiration
Frequent failures of electrical equipment such as transformers, motors, and generators often lead to power outages, costly repairs, and unplanned downtime. In many cases, these failures could have been prevented if early warning signs—like abnormal temperature or current—were detected in time.
As an engineering-focused team, we were inspired to combine core electrical engineering concepts with modern AI tools to move from reactive maintenance to predictive maintenance. Our goal was to build a simple yet impactful solution that engineers and utilities can realistically adopt.
What We Learned Through this project, we learned: How AI can enhance traditional engineering systems, not replace them Basics of predictive maintenance and fault analysis How to use the Google Gemini API for intelligent trend analysis and explanations Integrating machine learning models with web dashboards Using Firebase for backend services and data storage Turning raw sensor data into actionable insights This project strengthened our understanding of how AI can be applied responsibly in industrial and energy domains.
How We Built the Project Our system works in four main stages:
Data Input Users upload or enter electrical parameters such as: Voltage Current Temperature Load We used simulated CSV sensor data to represent real-world equipment readings. Fault Prediction (Machine Learning) A simple ML model analyzes the input data and predicts the risk level: Normal Warning Critical The idea is to keep the model lightweight, explainable, and practical.
AI Analysis with Gemini The prediction and data trends are sent to the Google Gemini API, which: Analyzes abnormal patterns Generates human-readable fault explanations Suggests preventive maintenance actions
Example insight: “The temperature rise combined with high current indicates potential overheating due to overloading.” Dashboard & Storage Results are shown on a clean web dashboard Data and results are stored using Firebase Firestore The app is hosted using Firebase Hosting
Technologies Used Google Gemini API – AI explanations and trend analysis Firebase & Firestore – Backend and database HTML, CSS, JavaScript – Frontend dashboard Python (scikit-learn) – Machine learning model
Challenges We Faced Designing a solution that is simple yet meaningful for a hackathon Creating realistic fault scenarios using simulated data Balancing engineering accuracy with AI-generated explanations Ensuring AI outputs are clear, responsible, and interpretable Integrating multiple technologies within limited time Each challenge helped us better understand real-world system design and teamwork under pressure. Future Improvements Real-time IoT sensor integration Mobile alerts (SMS / email) Support for smart grids and substations Advanced anomaly detection models
Conclusion This project demonstrates how AI and engineering can work together to make energy systems more reliable, efficient, and safe. By predicting failures before they happen, our solution aims to reduce downtime, save costs, and improve operational safety in the energy sector.
Log in or sign up for Devpost to join the conversation.