Real-Time Anomaly Detection in Manufacturing: A Data-Driven Approach to Enhancing Production Efficiency through Online KPI Monitoring

This study addresses the challenges of traditional value stream mapping in the context of Industry 4.0, where real-time data collection on the shop floor is becoming increasingly important. Traditional methods often fail to address the complexity of modern dynamic production environments leading to inaccuracies (lack of expert knowledge) and time delays (manual intervention) in root cause analysis. We propose an approach that uses online detection of KPI anomalies to proactively detect deviations in production processes and identify the root causes of productivity losses in real time. Our method combines statistical metrics, predictive techniques, and ensemble models to improve the robustness of anomaly detection. The system we use allows real-time monitoring with dynamic adaptation of the model to changing production conditions. Validated by case studies in automated and manual assembly processes, we show that the causes of productivity losses are reliably identified during operation in different production environments.