IOT-ENABLED PREDICTIVE MAINTENANCE FOR INDUSTRIAL AUTOMATION SYSTEMS

Abstract

Under Industry 4.0, the complexity of industrial automation systems is growing, making data-driven, intelligent maintenance methods necessary. In order to detect mechanical failures and pressure anomalies in a modular production line, this thesis investigates the creation of an Internet of Things (IoT)-enabled predictive maintenance (PdM) framework. In order to perform binary classification of mechanical failures and multi-class classification of pneumatic pressure levels, the system incorporates a multi-task neural network that uses sensor-generated time-series data. Node-RED is used for cloud logging and real-time data collection in the experimental setup, and feature extraction is carried out on scenarios that simulate both optimal and problematic operating situations. The model needs more work because, although it performs well in malfunction detection, it has trouble generalizing pressure predictions. Recent developments in PdM, such as the use of machine learning, deep learning, IIoT platforms, and explainable AI, are highlighted in an extensive literature analysis. This work lays the groundwork for future improvements in smart manufacturing systems by offering a scalable and reasonably priced PdM strategy that meets the demands of contemporary industrial automation.