Structural Anomaly Detection in Advanced Manufacturing Execution Systems

Anomaly detection is crucial to ensuring the robustness and reliability of operational systems. Early identification of anomalies and deviations from standard patterns prevents potential failures, enhances decision-making processes, and optimizes overall system performance. This paper investigates image-based structural anomaly detection for manufacturing execution systems utilizing an optimized VGG16 convolutional neural network to identify structural anomalies. The optimized VGG16 model significantly performs binary classification on the test data to distinguish normal and anomalous instances. A comparative analysis with another classifier demonstrates that the optimized VGG16 is notable with high anomaly detection accuracy and has the potential to improve system reliability significantly. Experimental findings on publicly available image-based anomaly datasets demonstrate the efficiency and effectiveness of the proposed technique in identifying anomalies in management execution systems.