PredictivePulse is an end-to-end Industrial IoT Predictive Maintenance Platform that combines Digital Twin simulation, MQTT messaging, TensorFlow-based machine learning, FastAPI services, MySQL data storage, Dockerized deployment, and a Flutter mobile application for real-time industrial monitoring.
The platform continuously simulates industrial machines, streams live telemetry through MQTT, predicts machine health using an LSTM model, estimates failure probability and Remaining Useful Life (RUL), stores historical sensor data, generates maintenance alerts, and visualizes the entire system through a live mobile dashboard.
PredictivePulse was developed as an industry-oriented Predictive Maintenance solution capable of simulating real industrial equipment and demonstrating a complete IIoT pipeline from sensor generation to AI-driven maintenance recommendations.
The project integrates multiple technologies into a single production-style architecture including:
- Digital Twin Simulation
- MQTT Communication
- TensorFlow Machine Learning
- FastAPI REST Services
- MySQL Database
- Docker Containers
- Flutter Mobile Dashboard
- Digital Twin Machine Simulator
- Multi-Machine Real-Time Simulation
- Live Sensor Data Generation
- Fault Injection Engine
- TensorFlow LSTM Health Prediction
- Failure Probability Estimation
- Remaining Useful Life (RUL)
- Maintenance Recommendation Engine
- FastAPI REST APIs
- MQTT Publisher & Subscriber
- MySQL Database Integration
- Historical Sensor Storage
- Alert Generation
- Swagger API Documentation
- Flutter Dashboard
- Live Fleet Monitoring
- Machine Details
- AI Predictions
- History Viewer
- Alerts
- System Status
- MQTT Live Updates
- Dockerized Deployment
- Mosquitto MQTT Broker
- MySQL Database
- Containerized Backend
- Python 3.11
- FastAPI
- SQLAlchemy
- Pydantic
- TensorFlow
- NumPy
- Pandas
- Scikit-Learn
- Joblib
- MySQL 8
- SQLAlchemy ORM
- MQTT
- Eclipse Mosquitto
- Flutter
- Provider
- MQTT Client
- HTTP REST API
- Docker
- Docker Compose
Digital Twin Simulator
│
▼
MQTT Publisher
│
▼
Mosquitto MQTT Broker
│
┌───────────────────┴───────────────────┐
│ │
▼ ▼
MQTT Subscriber Flutter App
│ (Live MQTT Updates)
│ ▲
▼ │
TensorFlow Prediction Engine │
│ │
▼ │
MySQL Database ───────────────► FastAPI REST APIs
│
▼
History • Alerts • Predictions
PredictiveMaintenance_V2/
│
├── backend/
│ ├── app/
│ │ ├── api/
│ │ ├── config/
│ │ ├── database/
│ │ ├── ml/
│ │ ├── mqtt/
│ │ ├── schemas/
│ │ ├── services/
│ │ └── simulator/
│ │
│ ├── .env.local
│ ├── .env.docker
│ └── Dockerfile
│
├── frontend_flutter/
│ ├── android/
│ ├── ios/
│ ├── lib/
│ ├── web/
│ ├── pubspec.yaml
│ └── analysis_options.yaml
│
├── artifacts/
├── data/
├── docs/
│
├── docker-compose.yml
├── requirements.txt
├── README.md
└── .gitignore
The project uses separate configurations for local development and Docker deployment.
backend/.env.local
Used by:
- MQTT Publisher
- MQTT Subscriber
backend/.env.docker
Used by:
- FastAPI Backend
git clone <repository-url>
cd PredictiveMaintenance_V2docker compose up -dThis starts:
- MySQL
- Mosquitto MQTT Broker
- FastAPI Backend
cd backend
python -m app.mqtt.subscribercd backend
python -m app.mqtt.publishercd frontend_flutter
flutter pub get
flutter runThe mobile application provides real-time monitoring of the industrial fleet.
- Splash Screen
- Fleet Dashboard
- Machine Details
- AI Predictions
- Sensor History
- Alerts
- System Status
Swagger UI
http://localhost:8000/docs
GET /
GET /prediction/{machine_id}
Returns
- Health Score
- Failure Probability
- Remaining Useful Life
- Recommendation
GET /history/{machine_id}
GET /alerts
Sensor Data
│
▼
Feature Engineering
│
▼
Feature Scaling
│
▼
TensorFlow LSTM Model
│
▼
Health Score Prediction
│
├────────► Failure Probability
├────────► Remaining Useful Life
└────────► Maintenance Recommendation
| Service | Port |
|---|---|
| FastAPI | 8000 |
| MySQL | 3306 |
| Mosquitto MQTT | 1883 |
| Mosquitto WebSocket | 9001 |
- machines
- sensor_data
- predictions
- alerts
Digital Twin Simulator
│
▼
MQTT Publisher
│
▼
Mosquitto Broker
│
▼
MQTT Subscriber
│
▼
TensorFlow Prediction Engine
│
▼
MySQL Database
│
▼
FastAPI REST API
│
├────────► Flutter Dashboard (REST)
│
└────────► History & Alerts APIs
MQTT Broker
│
▼
Flutter Live Dashboard (MQTT)
| Module | Status |
|---|---|
| Backend | ✅ Complete |
| Digital Twin Simulator | ✅ Complete |
| Machine Learning | ✅ Complete |
| MQTT Communication | ✅ Complete |
| FastAPI APIs | ✅ Complete |
| MySQL Database | ✅ Complete |
| Docker Deployment | ✅ Complete |
| Flutter Mobile App | ✅ Complete |
| End-to-End Integration | ✅ Complete |
- Angular Web Dashboard
- User Authentication
- Role-Based Access Control
- Push Notifications
- Predictive Maintenance Scheduling
- Cloud Deployment (AWS / Azure)
- Real Industrial PLC Integration
- CI/CD Pipeline
- Monitoring & Logging
- Conveyor Model Retraining
Sai Sanjanaa P R
B.E. Computer Science and Engineering (Internet of Things)
Sri Sairam Engineering College
Industrial IoT & AI Predictive Maintenance Internship Project
This project was developed as part of an Industrial IoT & AI internship to demonstrate an end-to-end Predictive Maintenance platform using modern software engineering, machine learning, messaging, and mobile technologies.