Predictive Maintenance

Submission for the course "Industrial Internet of Things" at Chaitanya Bharathi Institute of Technology, 2025.

Team Members

Name	Roll No
C. Rithesh Reddy	160122771034
G. Jayanth	160122771041

Introduction

This project is a predictive maintenance system that uses machine learning to predict equipment failures based on sensor data. The system consists of several components that work together to collect, process, and analyze sensor data in real-time. The system uses a Kafka broker to handle the data stream and an MQTT broker to collect data from sensors. The machine learning model is used to predict equipment failures based on the sensor data. The predictions are then sent to a Streamlit dashboard for visualization.

Architecture

Instructions to run the code

Requirements

• Python 3.8 or higher
• Kafka (3.3 or higher recommended)
• MQTT broker - Eclipse Mosquitto (2.0 or higher recommended)

Running the code

1. Clone the repository:

   git clone https://github.com/gjaynir0508/iiot-project.git

   This creates a directory named iiot-project in your current working directory.

   cd iiot-project

   Open this directory in your favorite code editor.

   code .

   Then open a terminal in the code editor or continue using the terminal you have open.

   The root directory of the project is the directory where the README.md file is located. All commands in this README file should be run from this directory.

2. Install the required packages:

   • Navigate to the project directory:

   • [Optional] Create a virtual environment:

   python -m venv venv

   • Activate the virtual environment:

     • On Windows: Run the following command in the terminal:

     venv\Scripts\activate

   • Install the required packages from the requirements.txt file:

   pip install -r requirements.txt

3. Start Kafka server

   • Follow the instructions in the Kafka Quickstart to start the Kafka server.

   • Create a topic named sensor_data:

   kafka-topics.sh --create --topic sensor_data --bootstrap-server localhost:9092 --partitions 1 --replication-factor 1

   • Create a topic named predictions:

   kafka-topics.sh --create --topic predictions --bootstrap-server localhost:9092 --partitions 1 --replication-factor 1

4. Start the MQTT broker

   • Add the mosquitto binaries to your PATH and run the following command in a terminal:

   mosquitto

   • There will be no output in the terminal, but the broker will be running in the background. You can check if the broker is running by opening a new terminal and running the following command:

   mosquitto_sub -h localhost -t test

   • If the broker is running, you will see no output in the terminal. You can stop the broker by pressing Ctrl + C.

5. Now we start running the project's components. Run the following commands in separate terminals in the same order, all in the root directory of the project:

   • Start the MQTT-Kafka bridge:

   python mqtt_kafka_bridge/bridge.py

   • Start the Inference server:

   python model/model_server.py

   • Start the Inference server:

   python model/model_server.py

   • Start the Prediction service that writes to the predictions topic:

   python model/predictor_kafka.py

   • Start the streamlit dashboard:

   streamlit run dashboard/streamlit_dashboard.py

Stopping the services

• To stop the services, press Ctrl + C in each terminal where the services are running. This will stop the services and close the terminals.
• Follow the reverse order in which they were started to stop the services. First, stop the streamlit dashboard, then the prediction service, then the inference server, and finally the MQTT-Kafka bridge.
• Finally, stop the Kafka server and the MQTT broker.