IoT- Enabled Manufacturing System with Freertos Using CAN Protocol for Predictive Maintenance

This paper presents an intelligent industrial automation system for predictive maintenance. This system combines real-time embedded control and cloud-based analytics to improve efficiency and availability. The real-time embedded components of the system use ESP32 microcontrollers to acquire data from sensors (temperature, voltage, current, and vibration) to monitor the health of a machine in real-time. The sensor signals from Sensor node are sent to the Central node transmitted via the CAN protocol using MCP2515 transceivers. The Central node implements a real-time operating system, FreeRTOS, allowing tasks to run concurrently to facilitate an efficient embedded processing. Sensor data is transmitted to the Azure IoT Hub using the MQTT protocol, which is lightweight for real-time transmission. A logistic regression model was trained in the Azure Machine Learning environment, and the algorithm was instantiated to perform binary classification on the incoming sensor data using supervised learning to predict the maintenance requirements and then predictive results are stored in a scalable SQL Server database for further analysis. Finally, the system integrates with Power BI to visualize both raw sensor data and predictive outputs through an interactive dashboard. The predictive capability helps in visualizing real-time machine health, current sensor data, and anticipated maintenance. The implemented system reliably handles sensor data, performs real-time embedded data processing, demonstrates high accuracy of predictive analytics and the cloud-based machine learning in a scalable solution for predictive maintenance in modern industry.