High impedance fault recognition in distribution system feeder

Dissertant

Jahangir, Haydar Kazim

Thesis advisor

Alias, Qais M.

University

University of Technology

Faculty

-

Department

Department of Electrical Engineering

University Country

Iraq

Degree

Master

Degree Date

2012

English Abstract

Power distribution feeders are prone to direct contact with neighboring objects, such as plant branches, such a contact constitutes a distribution feeder fault which called High Impedance Fault.

High Impedance Fault's are difficult to recognize and detect by traditional equipment because their presence results in slight increase in load current; thus can be confused with a normal increase in load.

A technique based on the combination of Fourier transform and artificial neural network is presented, simulated, and used.

Primarily the technique is used for the detection and recognition of high impedance fault in electric distribution power systems.

Moreover, it is used to detect and recognize system state changes.

The change in the line current and voltage signals caused by faults or any system switching event are decomposed to their harmonic components using the Fourier transform.

The signal features are extracted and the artificial neural network is trained using (Levenberg Back-Propagation Algorithm) to recognize these features.

Faults and system events investigated in this work are; High impedance fault (Line-toground, and line-to-line), low impedance fault (three-phase), normal load increase, nonlinear load switching, induction motor and shunt capacitor switching.

The verification and validation of the developed technique were accomplished via its application to two standard test systems.

The results obtained for these test systems were almost in exact match to those given in the literature [1, 6].

A comprehensive study is performed on a practical installed distribution feeder system.

This is a sector of Baghdad area distribution network called " Al-Khalij substation system".

All events mentioned above were performed with concluding remarks.

The results proved the ability of the technique and the algorithm developed to detect High Impedance Fault and Low Impedance Fault and identify them from other occurring events, such as; Normal load increasing, Nonlinear load, Induction motor starting, and Capacitor switching .

The cases studied and events were simulated and implemented using the SimPowerSystem Blockset in the MATLAB V7.10(2010a).

Main Subjects

Electronic engineering

Topics

• Neural networks(Computer science)

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-419999