Modeling and analysis of early-warning detection module in simulation lifecycle management framework using copula function

Abstract

The simulation lifecycle management framework is considered as an advanced manufacturing information system integrating product lifecycle management and manufacturing execution systems. While other manufacturing systems focus on the detections of current faults and the related controls, the framework has an early-warning detection module for predicting potential risks and for preventing them in advance. In order to design the module, the preliminary procedure is to construct the mapping model between manufacturing data and quality-based indicators. The mapping model is indicated as a nonlinear meta model. While neural network based models or response surface methods are applied for the meta model, it is limited in the fact that it is difficult to capture correlations among atypical manufacturing big data. In order to overcome the issue, a copula based nonlinear meta model is suggested with a numerical case study for clear understandings. The usage of copula theories helps to extract well-defined relationships among manufacturing data. The potential risks are predicted using the copular-based meta model and advanced controls are taken for preventing them effectively.