AI-Enabled Fault Detection for Predictive Maintenance of Ball Bearings

Manufacturing Industry 4.0 refers to the integration of advanced digital technologies and automation into the manufacturing sector. In the manufacturing industry 4.0, the Predictive Maintenance (PdM) plays a crucial role in optimizing equipment reliability, improving productivity, and reducing maintenance costs. The fault detection under predictive maintenance in ball bearing uses the sensor data and advanced analytics techniques to proactively identify faults, enabling timely maintenance actions and preventing equipment failures. The AI algorithms, such as Autoencoders, Artificial Neural Network (ANN), and Random Forest (RF) classification, enable the analysis of sensor data to detect and classify various types of bearing faults accurately making use of the Case Western Reserve University (CWRU) bench-marked dataset. The fault detection method is applied to ball bearings with fault diameters 0.007″, 0.014″, 0.021″ units. The Autoencoder model gets the greatest accuracy, outperforming the ANN and RF classification algorithms.