Parametric variations in dynamic models of induction machine clusters

This paper presents a probabilistic approach to the characterization of dynamical models of induction machine clusters. The authors' method derives bounds on eigenvalue variations for linearized models expressed in terms of stochastic norms. In their examples of the modeling of power system loads, this characterization tends to be less conservative than alternative deterministic approaches. They consider examples of induction machines with different ratings (classes), and allow for wide variations of electrical and mechanical parameters. They describe a stochastic norm approach to: (1) efficiently describe the dynamical model variations for a cluster of similar machines without having to perform repeated eigenvalue calculations, e.g. in a wind farm application; and (2) suggest the order of the reduced model in power system load modeling where the tightness of the bounds of eigenvalue variations is used for guidance in decisions regarding the number of different classes that would efficiently represent a given composite load.