Predictive Maintenance Approach in Textile Industry using Machine Learning

This study focuses on Predictive Maintenance approach in textile industry using Machine Learning (ML) and IoT to prevent sudden failures and improve operational efficiency. The model achieved 98% accuracy in predicting when the machine needs service, from real-time input provided by IR and vibration sensors The system continuously monitors four key parameters: Speed (RPM) using an IR sensor, Vibration (mm/sec) using a vibration sensor, Cloth Length (meters) entered manually, and Clip Depreciation (count) also entered manually. Real-time data from the IR and vibration sensors is collected using an ESP32 microcontroller, while the manually entered values are stored for analysis. The collected data is processed using the Random Forest algorithm, which classifies the machine’s condition as either normal or Time for Service. This classification helps in identifying early signs of wear and tear, ensuring timely maintenance before a major breakdown occurs. The processed results are stored and displayed in an Excel sheet, providing an easy-to-understand record of the machine’s condition over time. This predictive maintenance system is beneficial for textile industries as it reduces unexpected failures, lowers repair costs, increases machine lifespan, and minimizes downtime. By analyzing machine performance trends, industries can schedule maintenance efficiently, avoiding production delays and improving overall productivity. The system gets smarter through the integration of IoT and Machine Learning, allowing for real-time monitoring and data-driven decision-making. This ensures that cloth polishing machines operate smoothly, with fewer interruptions, ultimately leading to higher efficiency, better quality production, and cost savings for manufacturers.