Eunshin Byon, Ph.D.
Associate Professor, Depart of Industrial and Operations Engineering
University of Michigan
Abstract: Parameter Calibration in Large-scale Computer Experiments in the Era of Big Data
Advances in numerical algorithms and computing power bring simulation models to the forefront of design, control, interpretation, and analysis of many physical and engineered systems. Simulation models are typically developed based on physics-based first principles, and integrating simulation models with real-world data provides an enabling tool for understanding and optimizing system performance. Simulation models often require various parameter values to be appropriately specified. However, physical laws to accurately identify those parameters are often unavailable or insufficient in many applications. Parameter calibration is a procedure to identify those unknown parameters with observational data, aiming to improve the accuracy of simulation models and make them represent near-exact replicas of real systems. Typical calibration methods are based on surrogate modeling that imputes data under the assumption that physical and/or computer trials are computationally expensive. This, however, is not the case where large volumes of data can be collected during the operational stage. This research develops a quantitative scheme that provides an efficient and robust parameter calibration with Big Data when both observational data and simulation-generated data are not scarce. Despite the unprecedented opportunities that the abundance of data brings, making reliable decisions on the parameter calibration can still be time-consuming and computationally expensive even with advanced computing facilities. We develop novel approaches that effectively identify the best choice of subsets of data to guide the parameter calibration with theoretical and practical implications.
Department of Industrial & Systems Engineering