AI-Powered Predictive Maintenance System

Overview

AI-based predictive maintenance system built using Machine Learning, FastAPI, Spring Boot, Streamlit, and MySQL.

The project predicts industrial machine failure risk using real sensor data from the NASA CMAPSS turbofan engine dataset.

---

Features

• Real-time machine failure prediction
• FastAPI AI inference service
• Spring Boot backend integration
• Interactive Streamlit dashboard
• CSV bulk prediction
• Prediction analytics and charts
• MySQL prediction history storage
• Sensor health monitoring

---

Tech Stack

Frontend

• Streamlit
• Pandas
• NumPy

Backend

• Java
• Spring Boot
• REST APIs
• JPA/Hibernate

AI/ML

• Python
• FastAPI
• Scikit-learn
• Random Forest Classifier

Database

• MySQL

---

Architecture

Streamlit Dashboard ↓ Spring Boot Backend ↓ FastAPI ML Service ↓ Machine Learning Model ↓ MySQL Database

---

Dataset

NASA CMAPSS Turbofan Engine Dataset

---

Machine Learning Workflow

1. Load and preprocess NASA CMAPSS dataset
2. Train Random Forest Classifier
3. Save trained model using Joblib
4. Expose prediction API using FastAPI
5. Connect backend with Spring Boot
6. Visualize predictions in Streamlit dashboard

---

Features Implemented

• AI-based failure prediction
• Real-time REST API integration
• Sensor health analysis
• Prediction history
• CSV upload and bulk prediction
• Interactive dashboard and charts

---

How to Run Locally

Start FastAPI

cd ml-service
python -m uvicorn predict:app --reload

Start Spring Boot

cd spring-backend/backend
mvn spring-boot:run

Start Streamlit

cd streamlit-ui
python -m streamlit run app.py

---

Future Improvements

• Docker deployment
• JWT Authentication
• Kafka streaming
• React frontend
• Deep Learning (LSTM)
• Cloud deployment

---

Author

Shubham Kumar Jabalpur Engineering College