IFRPM

Intelligent Fleet Risk and Predictive Maintenance System

Project Structure

IFRPM/
├── backend/                    # Backend application (FastAPI)
│   ├── .env.example            # Environment variables template
│   └── app/
│       ├── config.py           # Application configuration
│       ├── database.py         # Database connection and session management
│       ├── main.py             # FastAPI application entry point
│       ├── seed.py             # Database seeding script
│       ├── ml/                 # Machine learning modules
│       │   ├── inference.py    # Model inference logic
        │       │   ├── loader.py       # Model loading utilities
│       │   └── stub.py         # ML stub implementations
│       ├── models/             # SQLAlchemy database models
│       │   ├── aircraft.py     # Aircraft model
│       │   ├── component.py    # Component model
│       │   └── rul_prediction.py  # RUL prediction model
│       ├── routers/            # API route handlers
│       │   ├── aircraft.py     # Aircraft endpoints
│       │   ├── alerts.py       # Alerts endpoints
│       │   ├── fleet.py        # Fleet management endpoints
│       │   ├── rul.py          # RUL prediction endpoints
│       │   └── weather.py      # Weather data endpoints
│       ├── schemas/            # Pydantic schemas for validation
│       │   ├── aircraft.py     # Aircraft schemas
│       │   ├── component.py    # Component schemas
│       │   └── rul.py          # RUL schemas
│       ├── services/           # Business logic services
│       │   ├── risk_service.py      # Risk assessment service
│       │   ├── rul_service.py       # RUL calculation service
│       │   └── weather_service.py   # Weather data service
│       └── utils/              # Utility functions
│           ├── feature_engineering.py  # Feature engineering
│           └── health_index.py        # Health index calculations
├── data/                       # Data storage
│   ├── processed/              # Processed data files
│   └── raw/                    # Raw data files
├── docs/                       # Documentation
│   ├── api.md                  # API documentation
│   └── backend.md              # Backend architecture documentation
├── frontend/                   # Frontend application
├── models/                     # Trained ML model files
├── notebooks/                  # Jupyter notebooks for analysis
└── requirements.txt            # Python dependencies

Predictive Maintenance Engine

The IFRPM backend inference engine supports a dynamic multi-model ensemble for calculating Remaining Useful Life (RUL).

By default, the application will scan the /models/ directory alongside the API and automatically ingest supported weights:

• Scikit-Learn Pickles (.pkl): Automatically loaded via joblib/pickle (e.g. ngafid.pkl, battery_xgb_model.pkl).
• Keras/TensorFlow Weights (.h5): Automatically compiled leveraging tensorflow.

The /api/rul and health endpoints will automatically compute an average of output predictions across your loaded .pkl and .h5 model nodes, safely bypassing missing expected components so dev environments won't crash when working iteratively.