A New Current Pattern Recognizing Protection

Technique for High Impedance Fault Using Cross Correlation

A. M. Sharaf, Senior Member, IEEE

University of New Brunswick,

Fredericton, NB, Canada

M. M. Eissa, Senior Member, IEEE

Faculty of Engineering, Helwan University,

EGYPT

High Impedance Arc-Type Faults (HIF) on Meshed Electrical Distribution/Utilization networks are characterized by an intermittent Arc-type nature and low-level of the fault currents.

Problem to be solve

In the multi-grounded distribution line, there exists unbalance in three phase loads, therefore, overcurrent ground relays are usually set high to allow some large neutral currents due to this unbalance.

The detection of low-level ground-currents using any conventional over-current or ground fault type relays is both difficult and sometimes inaccurate.

Each detection method may increase the possibility of the detection for high impedance faults to some extent, but it also has some drawbacks as well. Until now, no technique has offered a complete solution for this problem.

The paper presents the application of the cross correlation technique as a pattern recognition to high impedance faults (HIFs). The third and fifth harmonics current components are extracted from the fault current using Fast Fourier Transform (FFT).

Aim of the paper

Correlation is a measure of the relation between two or more variables. The measurement scales used should be at least interval scales, but other correlation coefficients are available to handle other types of data.

CROSS CORRELATION AS PATTERN CLASSIFICATION

Following the definition of the cross correlation function between x[n] and y[n] given by (1).

where k is a delay units.

(1)

The cross correlation functions of power signals are redefined such that the summations and integrations are replaced by averages. For two discrete power signals x[n] and y[n] the cross correlation function is defined as (2):

This feature of pattern classification is used in this paper.

(2)

The technique is based on a novel low frequency (the third and fifth harmonic feature diagnostic vector).

OVERALL PROCEDURE OF THE TECHNIQUE

The instantaneous current values at feeder substation buses shown in Fig. 1 are captured and transformed into frequency domain using one cycle Fast Fourier Transform FFT.

The FFT harmonic vectors extraction [i3] and [i5] are processed to obtain feature vectors.

The current pattern is classified using the cross correlation function given in (2).

S R

R1

F2 F1

X km

FFT

Pattern Classification

Fault Vector Feature

HIF with Linear or

Non-linear ARC

The slop of the Cross Correlation Function “XCF“ can be calculate to discriminate between linear and non-linear arc conditions

Threshold

• If the slop of “XCF“ goes lower than some THR_SLOP value, the technique will identify that the fault is linear HIF

• If the slop of “XCF“ goes higher than a THR_SLOP value, HIF is identified as a non-linear arc fault

The system includes a 138 kV. X is taken in per unit length. Data for verifying the proposed technique was generated by modeling the selected system using the Matlab/Simulink model

SYSTEM

The model

ic1v1

if

ic2

p1

is2

p2

1

den(s)

x lineiF

vF

i25

i23

i1

v1

v25

p5

p3

v23

i5

i3

v5

v3

p25

t

p1

p23

v2

i2

p2Source B

Source A

Scope

1

7e-3s+0.7

Rs Ls1

1

7e-3s+0.7

Rs Ls

In1Out1

NLL Linear/NL

In1Out1

LN

1s

1s

-K-

-K-

-K-

-K-

v2

i2

p2

v3v5i3i5

p3p5

FT1

v1

i1

p1

v3v5i3i5

p3p5

FT

-K-

1/(c*x)

-K-

1/(c*(l-x))

1

den(s)

(1-x) line

is1

i1

i2

The performance of the proposed technique has been evaluated for different types of internal faults. A wide variation of fault locations, source impedances, close in fault and fault resistances were investigated.

TEST RESULTS

Performance of the relay for a phase-a-to ground fault on the transmission line is shown in the following figure

The fault is located at 30% of transmission line length from R1

Effect of Internal Linear Fault

The corresponding computed “XCF“ for R1 has positive value. For the selected threshold boundary THR_SLOP, the “XCF“ slop is lower than this boundary. This indicates that the fault is a linear fault.

The computed [i3] and [i5] pattern is shown in Figure. The corresponding computed “XCF“ for R1 has very rise value 1.6E-02. For the selected threshold boundary, the “XCF“ does cross the selected threshold boundary

Effect of Internal Non-Linear Fault

The paper introduced a novel low order harmonic current pattern for HIGH IMPEDANCE FAULT ARC detection and discrimination.

CONCLUSIONS

The technique is based on analyzing the current pattern shape.

The suggested technique was tested under different fault conditions.

The paper introduced a novel low order harmonic current pattern for HIGH IMPEDANCE FAULT ARC detection and discrimination.

The great selectivity and reliability are the main features in discrimination between linear and non-linear arc faults.