AIoT Predictive Maintenance

Predicting machine failure before it happens — powered by machine learning

🔗 Live App

---

📖 About

AIoT Predictive Maintenance is a web app that predicts machine failure risk in real time from live sensor readings. It uses a Random Forest classifier trained with scikit-learn, achieving around 93% overall accuracy, to flag whether a machine is likely to fail based on its current operating conditions — helping catch problems before they cause downtime.

Feed it temperature, speed, torque, and tool wear readings, and it returns a prediction, a failure probability, and a plain-language status.

---

✨ Features

• 🎛️ Live sensor input form (machine type, temperature, speed, torque, tool wear)
• 🤖 Real-time failure prediction via a trained Random Forest model
• 📊 Failure probability score with a color-coded risk gauge
• 🚦 Status indicator: Normal / Watch / Alert
• ⚡ Fast, lightweight FastAPI backend

---

🛠️ Tech Stack

Layer	Technology
Frontend	HTML, CSS, JavaScript
Backend	FastAPI
Model	scikit-learn — Random Forest classifier
Dataset	AI4I 2020 Predictive Maintenance Dataset
Accuracy	~93%
Deployment	Render

---

📁 Project Structure

aiot-predictive-maintenance/
├── models/
│   ├── maintaince_model.pkl
│   └── preprocessor.pkl
├── static/
│   ├── script.js
│   └── style.css
├── templates/
│   └── index.html
├── Dockerfile
├── README.md
├── app.py
├── requirements.txt
└── schemas.py

---

🚀 Setup

1. Clone the repository

   git clone https://github.com/httpsanjay/iot-predictive-maintenance-model.git
   
   cd iot-predictive-maintenance-model

2. Create a virtual environment and install dependencies

   python -m venv venv
   source venv/bin/activate    
   pip install -r requirements.txt

3. Run the app

   uvicorn app:app --reload

4. Open http://localhost:8000 in your browser

⚠️ Note: The included Dockerfile isn't fully working yet — a fix is on the way. For now, run the app locally with the steps above, or use the live deployment linked at the top of this README.

---

🔌 API Reference

POST /predict

Request body

{
  "Type": "L",
  "Air temperature [K]": 298.1,
  "Process temperature [K]": 308.6,
  "Rotational speed [rpm]": 1500,
  "Torque [Nm]": 40.5,
  "Tool wear [min]": 120
}

Response

{
  "prediction": 0,
  "failure_probability": 34.1,
  "status": "No maintenance required"
}

Field	Type	Description
Type	string	Machine quality variant — L, M, or H
Air temperature [K]	float	Ambient air temperature
Process temperature [K]	float	Process temperature
Rotational speed [rpm]	int	Spindle rotational speed
Torque [Nm]	float	Applied torque
Tool wear [min]	int	Cumulative tool wear time
prediction	int (0/1)	1 if the model classifies this as a failure case
failure_probability	float (0–100)	Model's confidence that the machine will fail
status	string	Human-readable status shown in the UI

---

License

Licensed under the MIT License.

---

Made with 🩷 by Sanjay