Sensitivity, approximation, and uncertainty in power system dynamic simulation

Parameters of power system models, in particular load models, are seldom known exactly, yet dynamic security assessment relies upon simulation of those uncertain models. This paper proposes a computationally feasible approach to assessing the influence of uncertainty in simulations of power system dynamic behavior. It is shown that trajectory sensitivities can be used to generate accurate first-order approximations of trajectories that arise from perturbed parameter sets. The computational cost of obtaining the sensitivities and perturbed trajectories is minimal. The mathematical structure of the trajectory approximations allows the effects of uncertainty to be quantified and visualized using worst-case analysis and probabilistic approaches.