GRIDGUARD AI

Autonomous grid resilience system that detects anomalies, predicts failures, and protects critical infrastructure.

Comment

Inspiration

Power grid failures cost billions of dollars and cost lives. The 2021 Texas power crisis killed 246 people and caused $195 billion in damages. Hospitals lost power, emergency services failed, and people froze in their homes. Current monitoring systems only react after failures happen - they don't predict or prevent. GRIDGUARD was built to change that.

What it does

GRIDGUARD AI is an autonomous grid resilience system that:

  • Detects abnormal energy conditions (voltage drops, current spikes, overheating)
  • Predicts instability with 91%+ confidence
  • Protects critical infrastructure automatically (hospitals, emergency services, communications)
  • Simulates interventions before executing them - reducing risk from 100% to 43%

The system features a professional dashboard with real-time data, a predictive risk engine, an autonomous agent, resilience simulation, and critical infrastructure protection.

How I built it

Frontend: HTML5, CSS3, JavaScript
Backend: Python, Flask
Deployment: Vercel, GitHub Pages
Other: REST API, JSON

The frontend dashboard displays live grid data with 6 interactive cards. The backend provides REST API endpoints and serves the frontend. The system generates realistic simulated grid data with daily usage patterns and fault conditions.

Challenges I ran into

Not having access to real hardware sensors. Since I couldn't use physical sensors like temperature monitors, voltage meters, or current sensors, I had to build a realistic simulation engine that generates believable grid data. I spent significant time making the simulated data look and behave like real-world energy readings—with daily usage patterns, random fluctuations, and realistic fault conditions.

Deploying the backend to Vercel was also challenging because Vercel doesn't support WebSocket connections. I had to redesign the real-time updates to work with HTTP polling instead.

Accomplishments that I'm proud of

  • Building a complete full-stack application from scratch
  • Deploying a working live demo: https://gridguard-ai-v2.vercel.app/
  • Creating a professional UI with clean white/brown theme
  • Achieving 91%+ confidence in risk predictions
  • Designing a simulation-first approach that reduces risk from 100% to 43%
  • Implementing keyboard shortcuts (1-6) for quick scenario switching

What I learned

  • Full-stack development with Flask and vanilla JavaScript
  • REST API design and implementation
  • Deployment to Vercel and GitHub Pages
  • How to build realistic simulation engines
  • Handling serverless limitations (replacing WebSocket with polling)
  • Creating professional UI without frameworks

What's next for GRIDGUARD AI

  • Real sensor integration (ESP32, Arduino)
  • Machine learning model training with real grid data
  • Mobile application
  • SCADA system integration
  • Cloud deployment (AWS, Azure)
  • Email and SMS alert system
  • Data export and reporting

Live Demo: https://gridguard-ai-v2.vercel.app/
GitHub: https://github.com/yanetbedilu7-00/gridguard-ai

Built With

Share this project:

Updates

Submission history