CAREER: Intelligent Information Processing for Power Systems

Drexel University, Philadelphia PA

Investigators

Abstract

This proposal the question of how an electric power utility can extract information about the composition of the instantaneous load without having to introduce a multitude of additional measurements is studied. A statistical method to be employed allows to separate individual loads out of the aggregate of existing measurements. These measurements are usually superposition of different types of loads and generators, as obtained, for instance, at a substation level. The method called "Blind Source Separation" or "Independent Component Analysis" (ICA) requires only minimal assumptions, like independence of loads and linear superposition of currents, which can be assumed due to the basic properties of load distribution in realistic power networks. ICA has been applied successfully for the solution of technical problems previously, and, although it has never been employed for power systems, preliminary data given indicate that the method provides a powerful and reliable way of achieving load profile separation and identification. It is also conjectured that this method is suitable for harmonic injection identification. More generally, it is suggested that ICA may become a powerful method to achieve dimensionality reduction of data in a multitude of applications in power system analysis, operations and control.