The nonlinear nonsmooth nature of power system dynamics complicates the pro
cess of validating system models from disturbance measurements. This paper
uses a Gauss-Newton method to compute a set of model parameters that provid
e the best fit between measurements and model response. Trajectory sensitiv
ities are used to identify parameters that can be reliably estimated from a
vailable measurements. An overview of trajectory sensitivity analysis is pr
ovided. An example based on the Nordel system is used throughout the paper
to illustrate the various concepts. This example exhibits both soft and har
d nonlinearities.