National Institute of Standards and Technology
Modeling methodology is a vast subject, encompassing empirical and analytical means of discovery and numerical techniques. Design engineering and operations research are two areas where modeling can be key to success. In these two areas alone, the great variety of questions that could be explored, and methods applied, might make one wonder whether there could be, nonetheless, general means to integrate knowledge from models directly with other enterprise knowledge. A question that arises early is whether there might be value in treating equations as information objects, in the sense that system models, component data sheets, and production data are information objects, managed and interrelated by various software applications. Could exposing equations in this sense enhance knowledge refinement and reuse? Could it facilitate model verification? Agility? In this presentation I will discuss how NIST’s Modeling Methodology for Smart Manufacturing project is approaching these questions. We are using metamodels and semantic web technology to link equations into wider information infrastructure and support their independence. I will describe an investigation in which equations representing predictive models of a unit manufacturing process were developed and interrelated with production-oriented information. In on-going work, we are using mapping technology, templates, and trade-off analysis to formulate optimization problems, providing optimal process parameters.
Peter Denno is a computer scientist in the Systems Integration Division. He has 32 years’ experience in manufacturing research including 20 years at NIST and 10 years at Pratt & Whitney. His first encounter with metamodeling was 19 years ago, when he developed the Expresso EXPRESS validation system. He has remained involved with the technology since that time.