Data Driven Process Flow Modeling and Optimization Using Indoor RFID Locating System




Chang, Chi-Wen

Journal Title

Journal ISSN

Volume Title



The comprehensive research delves into the multifaceted realm of data science, leveraging the capabilities of data-driven process flow modeling and optimization. The collective findings underscore the potential of data-driven methodologies and the indoor Radio Frequency Identification (RFID) locating system in diverse domains. From enhanced data integrity and predictive modeling to revolutionary healthcare monitoring and advanced event recognition, this comprehensive study charts a path toward innovation and optimization. Pixel matrix charts serve as a representation tool, considering both spatial and temporal parameters to visualize equipment paths. The fusion of automation, machine learning, and Lean Six Sigma principles provides a comprehensive framework for addressing complex challenges in data-driven environments, making this research an invaluable contribution to the field. The major contribution of this research offers a unique perspective on anomaly detection in the context of medical equipment flow, particularly under uncertain conditions. Extensive analysis culminates in the development of a multi-step process that seamlessly integrates Convolutional Neural Networks (CNN) and Prototypical Neural Networks (PNN). This innovative methodology elucidates the temporal-spatial flow paths of medical equipment in healthcare facilities, enabling the identification of equipment travel patterns amid various uncertainties. Experimentally, the integration of CNN feature extraction with PNN exhibited an outstanding accuracy rate of 99.97% for classification tasks, marking a significant contribution to the development of anomaly detection mechanisms.



Radio Frequency Identification, Prototypical Neural Networks, Convolutional Neural Networks, Data-driven process



Mechanical Engineering