Abstract
Realistic mathematical modeling is critical in engineering. Models help generalize experimental results, often suggest new experiments, and can replace costly physical testing. Most models, however, require validation of their parameters with experiment. For some models, parameters can be directly measured, but others must be inferred. In both cases parameters selection should reflect the inherent uncertainty in both measurement and modeling. This is particularly important for models used in control algorithms in which unexpected inputs may delay or prevent appropriate response. In this work, a method of characterizing uncertainty in model parameters is presented, the steps of the method are described in detail and illustrated using a case study application which models the force produced by a magneto-rheological (MR) damper. In this probabilistic approach, the model parameters are treated as random variables. The Metropolis-Hasting algorithm is used to generate sample parameter sets. These sets provide a probabilistic characterization of the parameters, can illustrate potential correlations between parameters and may offer insight into the effect of uncertainty in algorithms used for control applications. (C) 2016 American Society of Civil Engineers.