Murdoch University Research Repository

Welcome to the Murdoch University Research Repository

The Murdoch University Research Repository is an open access digital collection of research
created by Murdoch University staff, researchers and postgraduate students.

Learn more

The use of artificial intelligent in power system stability

Kharrazi, Ali (2015) The use of artificial intelligent in power system stability. Other thesis, Murdoch University.

PDF - Whole Thesis
Download (2MB)


In this project the use of Artificial Intelligence in Power System Stability Enhancement is studied. The focus of the project is on low frequency oscillation in power systems. These oscillations known as electromechanical modes of oscillation were observed in power systems in the early 1960’s. They are the result of kinetic energy exchange between synchronous generators in a power system. Power System Stabilizers has been used to dampen these oscillations. Conventional Power System Stabilizers (CPSS) have been used for decades in power systems. They consist of a washout filter and two lead lag blocks to compensate the phase lag of the system.

Tuning of CPSS is of paramount importance for proper response and function. In this project, a Genetic Algorithm (GA) optimization process is introduced in order to tune the parameters of a CPSS in a Single Machine Infinite Bus system. The power system is modelled in PowerFactory and GA optimization uses the MATLAB Optimization Toolbox. An effective data exchange is engineered between PowerFactory and MATLAB to facilitate the tuning process. The optimization is performed using the position of the eigenvalues of the power system which are found by running Modal Analysis in PowerFactory. The main goal of optimization is to tune the parameters of CPSS to maximize the dampening ratio of all oscillatory modes.

In the next section of the project an Artificial Neural Network (ANN) based power stabilizer is introduced. Two neural networks are designed. First is a Neural Identifier to model the dynamics of the system and to predict the future output of the system. Second is a Neural Controller to provide proper stabilizing signals to dampen the oscillation by comparing the output of Neural Identifier with the desired value. A data interface between PowerFactory and MATLAB is established in order to enable the training and simulation of the proposed control structure.

Item Type: Thesis (Other)
Murdoch Affiliation(s): School of Engineering and Information Technology
Supervisor(s): Cole, Graeme
Item Control Page Item Control Page


Downloads per month over past year