SealGuard is an AI-driven monitoring system designed to detect moisture ingress and air leaks in oil-filled power transformers. By analyzing the relationship between Oil Temperature (breathing) and Silica Gel Breather Bubbles, the system uses an Isolation Forest algorithm to identify anomalies in real-time.
- Real-Time Simulation: Simulates realistic transformer heating and cooling cycles using a virtual ESP32.
- AI Anomaly Detection: Uses an Isolation Forest (Machine Learning) model to detect leaks.
- Interactive Dashboard: Built with Streamlit to visualize live telemetry data.
- Dual-Axis Telemetry: Tracks Temperature vs. Bubble Count on a synchronized timeline.
- Smart Guardrails: Hybrid logic (AI + Rules) ensures zero false alarms during heating cycles.
- Language: Python 3.x
- Dashboard: Streamlit
- ML Model: Scikit-Learn (Isolation Forest)
- Visualization: Altair
- Data Processing: Pandas, NumPy
- Simulation: Custom Python Physics Engine (
fake_esp.py)
| File | Description |
|---|---|
fake_esp.py |
The Hardware Simulator. Simulates the ESP32, Temp Sensor, and Bubble Counter. |
dashboard.py |
The UI. A Streamlit app that visualizes data and displays AI status. |
generate_data.py |
The Teacher. Generates 50,000 steps of synthetic "Healthy" data. |
train_model.py |
The Brain. Trains the Isolation Forest model and saves it as .pkl. |
sealguard_model.pkl |
The saved Machine Learning model (generated by train_model.py). |
esp_data.txt |
Live data buffer shared between the simulator and the dashboard. |
Clone the repository and install the required Python packages:
git clone [https://github.com/Rithwik-Ravi/SealGuard.git](https://github.com/Rithwik-Ravi/SealGuard.git)
cd SealGuard
pip install pandas streamlit scikit-learn altair joblib