Cyber-enabled Manufacturing Systems (CeMS)

Abstract:

Quality control in high-volume manufacturing is commonly performed using Statistics-based quality control techniques. These techniques require large data sets in order to specify acceptable variation limits. These statistics-based QC approaches are not applicable in small-lot high-value manufacturing where it is important to ensure the quality of each one of the final products. Previous model based process control used simplified representations of the actual physics of the process. While high-dimensional nonlinear models have often been used as offline tools for detailed study of manufacturing processes and for optimization, but their high computational cost had prevented their application to real-time control. Cyber-Enabled Manufacturing System is an intelligent manufacturing approach that incorporates high-fidelity physics-based models with multiple, but uncertain, measurements of a manufacturing process into a computer-driven monitoring and control system. The control deadline combined with the high computational cost of high-fidelity models requires a deeper understanding of the computations, how they are carried out, and how they affect the physical system. The main objective of creating a CeMS is to perform real-time monitoring and prediction of manufacturing defects. This advancement means changes in manufacturing parameters such as substitution of materials, differing processing speeds, differing product geometries, etc., can be handled by the model-based controller in real-time without new extensive testing as required with present statistical control. CeMS require the ability to accurately model the physical system in order to predict its future state based on the estimated current state and inputs. The model must be simulated faster than real time to be used to monitor the formation of defects. Cyber-physical systems (CPS) focus on optimizing the interaction of computing with the evolving state of the physical system, in order to establish a defect-free environment for the target manufacturing application. In traditional process control systems, computations for control and estimation are assumed to be completely independent of the physical system. In CPS the nature and amount of computation may depend on the state of the physical component. In our proposed approach, the computational process works in close synchronization with the physical process, resulting in collaboration of both the physical and cyber components.