Continuous Process Monitoring Through Ensemble-Based Anomaly Detection

Abstract

In many production processes, a complete quality inspection of all products is not feasible due to technological and organizational restrictions. In order to ensure zero-defect products, monitoring process parameters in real time and using them to predict product quality by supervised learning methods is a very established approached. However, this approach requires a joining of process parameters and quality features. In order to guarantee high-quality products even in the absence of traceability, a continuous process monitoring approach based on an anomaly detection ensemble method is beneficial.