This repository is forked from Mayank8159/jet_engine as part of a team hackathon project.

AeroGuard: Jet Engine Predictive Maintenance

AeroGuard is an AI-powered intelligence terminal designed for real-time monitoring and Remaining Useful Life (RUL) prediction of jet engines. By leveraging a Deep Learning LSTM (Long Short-Term Memory) model, the system processes multi-dimensional sensor telemetry to predict engine failure before it occurs.

🛠 Project Architecture & Workflow

The system is split into two primary layers: the Inference Engine (Python) and the Intelligence Terminal (Next.js).

1. Data Processing

• Input: 24 sensor channels (Temperatures, Pressures, Fan Speeds) over a 30-cycle time window.
• Normalization: Data is passed through a StandardScaler to match the training distribution of the CMAPSS dataset.

2. Backend (FastAPI + TensorFlow)

• Inference: The LSTM model analyzes temporal patterns in sensor degradation.
• Decision Logic: The RUL output is converted into a Health Index (%) and a Status Grade (A/B/C).
• REST API: Exposes a /predict endpoint that accepts 30x24 matrices.

3. Frontend (Next.js 15 + Tailwind CSS)

• Fleet View: A macro-level dashboard for managing multiple assets and triaging critical units.
• Single Engine View: A micro-level deep dive allowing manual telemetry input and sensor impact analysis.

---

🚀 Getting Started

Backend Setup (Python)

1. Navigate to the /backend directory.
2. Install dependencies:

pip install fastapi uvicorn tensorflow joblib numpy

3. Start the server:

uvicorn main:app --reload --port 8000

Frontend Setup (Next.js)

1. Navigate to the /frontend directory.
2. Install dependencies:

npm install

3. Run the development server:

npm run dev

4. Access the dashboard at http://localhost:3000.

---

📊 System Workflow

Step	Action	Description
1	Telemetry Ingest	30 time-steps of 24 sensor values are collected.
2	JSON POST	Frontend sends data to localhost:8000/predict.
3	LSTM Inference	Model predicts RUL (cycles remaining).
4	Risk Mapping	Backend calculates Status (Healthy/Warning/Critical).
5	Visualization	Dashboard renders degradation trends and impact scores.

---

📁 Key Components

• main.py: The FastAPI server handling model loading and inference logic.
• SingleEngine.tsx: Interactive component for deep-dive analysis.
• FleetDashboard.tsx: Grid-based overview for operational triage.
• parseCSV.ts: Strict data cleaning utility to ensure 30x24 shape validation.

---

⚠️ Requirements

• Python 3.9+
• Node.js 18+
• Model Assets: Ensure lstm_rul_model.h5 and scaler.pkl are in the backend/models/ folder.