International Journal of Computational
Intelligence Research (IJCIR)
Volume 2, Number 1 (2006)
Hierarchical neural networks toward a unified modeling framework for load dynamics
Siemens Power Transmission and Distribution, Inc., Minnetonka, MN 55305, USA
Tsinghua University, Department of Automation, Beijing 100084, China
Ronald R. Mohler
Oregon State University, Department of Electrical and Computer Engineering, Corvallis, OR 97331, USA
Hierarchical neural networks have found their use in control applications such as intelligent control design, identification, pattern classification, etc. In this paper, modeling of load dynamics is studied within a proposed unified framework that is general enough so that it can be applied to voltage stability analysis, load forecast, real time load prediction, etc. The study is conducted in conjunction with the application of hierarchical neural networks. The properties of hierarchical neural networks are studied to provide the theoretical justification for the application of hierarchical neural networks to load dynamics modeling within the proposed unified framework. It is shown that modeling of load dynamics can be formulated in such a way that application of hierarchical neural networks becomes naturally logical. A few theoretical results on hierarchical neural networks for load modeling are presented. Furthermore, a study case is presented to illustrate how hierarchical neural networks can be applied and how they perform. This demonstrates the effectiveness of the proposed unified framework for modeling of load dynamics.
Neural Network, Hierarchical Neural Network, Load Dynamics, Dynamic Load Modeling, Voltage Stability Analysis, Load Forecast.