Adaptive Predictive Maintenance Models for Machine Failures Detection and Diagnosis

Predictive maintenance (PM) is essential for the manufacturing industry. Mainly, due to its complicated operations, there is a great need for the semiconductor industry to increase efficiency. In this presentation, two adaptive frameworks, including predicting machine health conditions and predictive production monitoring, will be presented. It employs deep adaptive learning for online machine anomaly detection to address dynamic changes in environmental conditions. Additionally, it is diagnosing novel machine failures using transfer learning. Additionally, we considered reliability metrics for maintenance, such as Mean Time to Repair. These analytics are used for predicting future production yields to help manufacturers avoid unsatisfactory production levels.