Fault Detection Techniques for Process Parameter Monitoring in Dairy Milk and Cream Fat In-Line Standardization: A Comparative Analysis

Sammanfattning

Detecting and recognizing when there is a deviation in process behavior is a crucial aspect of industrial automation. This study investigates one such deviation, a volume discrepancy accidentally introduced during the commissioning of a Tetra Pak Standardization Unit (TPSU). The analysis aims to develop theory for a future implementation with the purpose of detecting this deviation by evaluating principal component analysis-based (PCA-based) detection methods and frequency spectrum analysis (FSA). The comparison focuses on the accuracy, validity, reliability, and feasibility of the detection methods. Results demonstrate that FSA is the most promising method and two approaches to implementation are proposed, evaluating the strengths and limitations of each approach. This work acts as a guide and framework for Tetra Pak to use during a potential future implementation and provides insights into necessary future work.