ELK415E COMPUTER RELAYING

FOR POWER SYSTEMS

CHAPTER 2.3: Digital Signal Processing

for Computer Relaying
The purpose of this chapter is to give an elementary
grounding in digital signal processing including the
process and limitations of sampling, digital filtering and
spectral analysis.

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 1

Continuous and discrete-time waveforms
As the names suggest, this classification is determined by
whether or not the time axis is discrete (countable)
or continuous. A continuous-time signal will contain a value
for all real numbers along the time axis.

In contrast to this, a discrete-time signal, often created by

sampling a continuous signal, will only have values at
equally spaced intervals along the time axis.
If a 50Hz waveform is displayed on a standard oscilloscope,
then the trace of the waveform is said to be continuous, that
is, at every point in time, there is a distinct value which
represents the 50Hz waveform. All power system waveforms
are continuous, as are the waveforms from a microphone or a
record player.

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 2

Continuous and Discrete-Time Signals
Most of the signals we will talk about are functions of time.
There are many ways to classify signals. This class is organized
according to whether the signals are continuous in time, or
discrete.
A continuous-time signal has values for all points in time in some
(possibly infinite) interval.
A discrete time signal has values for only discrete points in time.

Continuous-Time Signal Discrete-Time Signal

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 3

 f(t) = f(T+t)

A periodic signal with period To

f(t) ≠ f(T+t)
An aperiodic signal

As the name implies, signals can be characterized as to

whether they have a finite or infinite length set of values.
Most finite length signals are used when dealing with
discrete-time signals or a given sequence of values.
Mathematically speaking, f(t) is a finite-length signal if it
is nonzero over a finite interval t1<f(t)<t2,
where t1>−∞ and t2<∞
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 4
 An example can be seen in Similarly, an infinite-length signal, f(t),
is defined as nonzero over all real numbers: ∞≤f(t)≤−∞

Causal vs. Anticausal vs. Noncausal Signals

Causal signals are signals that are zero for all negative time,
while anti-causal are signals that are zero for all positive
time.

Noncausal signals are signals that have nonzero values in

both positive and negative time

A causal signal An anticausal signal A noncausal signal

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 5

Even vs. Odd Signals
An even signal is any signal f such that f(t)=f(−t). Even signals
can be easily spotted as they are symmetric around the
vertical axis. An odd signal, on the other hand, is a
signal f such that f(t)=−f(−t)

An even signal
An odd signal
Consider the signal defined for all real t described by

f(t)={sin(2πt)/t for t≥1 and 0 for t<1

This signal is continuous time, analog, aperiodic, infinite length,
causal, neither even nor odd, and, by definition, deterministic.
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 6
 Deterministic vs. Random Signal
A deterministic signal is a signal in which each value of the
signal is fixed and can be determined by a mathematical
expression, rule, or table. Because of this the future values of
the signal can be calculated from past values with complete
confidence.
On the other hand, a random signal has a lot of uncertainty
about its behavior. The future values of a random signal
cannot be accurately predicted and can usually only be
guessed based on the averages of sets of signals

Deterministic Signal Random Signal

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 7
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 8
Types of Systems
Systems are classied according to the types of input and output
signals
• Continuous-time system has continuous-time inputs and
outputs.
AM or FM radio
Conventional (all mechanical) car
• Discrete-time system has discrete-time inputs and outputs.
PC computer game
Matlab

• Hybrid systems are also very important (A/D, D/A converters).

You playing a game on a PC
Modern cars with ECU (electronic control units)
Most commercial and military aircraft
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 9
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 10
Discrete-Time Signals

X[n]= X(n Δt)

n is the sampling instant,
x[n] is the value at the sampling instant
{x[n]} represents the sequence

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 11

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 12
Time Scaling

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 13

Time Reversal

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 14

Time Reversal

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 15

Time scaling, shifting, and reversal can all be combined.
Operation can be performed in any order, but care is required.
Example: x(2(t - 1))
Scale First, then shift, Compress by 2, shift by 1

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 16

 Linear Time Invariant Systems

A discrete-time signal such as x[n] or y[n] is described by an

infinite sequence of values, i.e., the time index n takes values in
-∞ to +∞.
In the diagram above, the sequence of output values y[.] is the
response of system S to the input sequence x[.]

The system is causal if y[k] depends only on x[j] for j≤k

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 17

Time Invariant Systems

Let y[n] be the response of S to input x[n]. If for all possible

sequences x[n] and integers N

then system S is said to be time invariant (TI). A time shift in

the input sequence to S results in an identical time shift of the
output sequence.

In particular, for a TI system, a shifted unit sample function

δ[n-N] at the input generates an identically shifted unit sample
response h[n-N] at the output

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 18

 Linear Systems

Systems Let y1[n] be the response of S to an arbitrary input

x1[n] and y2[n] be the response to an arbitrary x2[n]. If, for
arbitrary scalar coefficients a and b, we have:

then system S is said to be linear. If the input is the weighted

sum of several signals, the response is the superposition (i.e.,
same weighted sum) of the response to those signals.

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 19

Unit Sample and Unit Step Responses of Linear Time-
Invarian System- LTIS

LTIS
δ[n] = u[n]− u[n −1] h[n] = s[n]− s[n −1]

The unit sample response of a system LTIS is the response of

the system to the unit sample input. We will always denote the
unit sample response as h[n].

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 20

 Similarly, the unit step response s[n]:

1 n  0
u[n]  
0 n  0 LTIS
{X[n]} = u[n] Unit step sequence s[n] = h[k]

u[n]   [n]   [n  1]   [n  2]     [n  k ]
k 0

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 21

Relating h[n] and s[n] of an LTI System

Then

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 22

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 23
A discrete-time signal can be decomposed into a sum of time-
shifted, scaled unit samples.

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 24

 Convolution
Convolution is the process by which an input interacts with an
LTI system to produce an output Convolution between of an
input signal x[n] with a system having impulse response h[n] is
given as,

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 25

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 26
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 27
Digital filtering
The sequence of values of the sampled waveform x is described by
x(nT), where T=Δt is the sampling interval and n is an index which
allows any value in the sequence to be specified.
Commonly, T or Δt omitted in this description since it is always
implied and, the description of the sequence is simply given as x(n).
The original continuous waveform is described as x(t),where t is the
continuous time variable.
After sampling, the input signals to a digital protection relay
may contain frequencies at other than 50Hzwhich will need to
be removed to ensure correct relay operation since, as stated
earlier, digital relaying algorithms are usually based on 50Hz
signals. The most efficient way of filtering out non-50Hz
components is by the use of a digital filter.

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 28

 Digital filtering
Digital filtering process operates on the sampled values of
power system waveforms.
Figure shows a waveform which is clearly not a sine wave, in
actual fact it is the result of superimposing the following sine
waves:
(i) 50 Hz sine wave, magnitude 1
(ii) 50 Hz sine wave, magnitude 1/3
(iii) 250 Hz sine wave, magnitude 1/5

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 29

 Fourier
transform

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 30

Digital filtering
Digital Filter y[n]
X[n]
Input sequence Output sequence

Unit impulse or Unit impulse response

Unit sample or Unit sample response

The impulse response of a filter is as unique as its frequency response

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 31
Digital filtering
In general, the output of a digital filter is the summation of the
individual responses of the filter to each sample in the input
sequence. Figure shows this in more detail where a digital filter, with
specified impulse response, has a sine wave based sequence of
samples as its input.

x[n] LTIS y[n]

δ[n] Digital Filter h[n]

X[n] y[n]
h[n]

Input sequence Output sequence

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 32

Graphical explanation of
digital convolution

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 33

Let the sequence of values of the impulse response be denoted by h[k]
for k = 1,2,3,4 and 5. Note that the values of h[k] are usually referred
to as the filter coefficients.

Let the input sequence values be denoted by x[n] for n = 0,1,2, 3 ......
and the filter output sequence be denoted by y[n] for n = 0,1,2, 3 .....
At sample n=5, the output of the filter can be expressed as:

y[5] = x[5] × h[1] + x[4] x h[2] + x[3] x h[3] +x[2] x h[4] +x[l] x h[5]

In general, the output of the filter at sample n is given by:

y[n] =x[n] × h[1] +x[n - 1] × h[2] +x[n - 2] × h[3] + x[n - 3] × h[4]
+ x[n - 4] x h[5]

This is essentially an equation for digital convolution and may be

written in a more convenient form using the mathematical summation
sign:
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 34
Digital filtering

A more general expression for digital convolution is expressed as

Where, N is the number of impulse

response coefficients

A filter of this type is, thus, referred to as a finite impulse

response (FIR) filter. FIR filters have the property that
their group delays are never greater than NT where T is
the sampling interval. Furthermore, when designing FIR
filters, the group delay may be used as a design parameter.

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 35

Another important class of digital filters have an infinite impulse
response (IIR). IIR filters cannot be implemented via above
Equation. Instead an equation which uses both past values of the
filter input and output is used:

Where, a and b are a set of M filter coefficients. Since previous

values of the filter output are used in above Equation, this equation
is described as being recursive.

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 36

 Discrete Fourier transform
The process of moving between the time domain and the frequency
domain, or vice-versa, is referred to as Fourier transforming.
Quite often, the impetus for performing Fourier transforms is the need
to have a knowledge of the frequency spectrum of some time domain
waveform, hence the section rifle, spectral analysis.

In practice, the DFT, for transforming from rime to frequency

domains, is implemented as two equations:

It is used to estimate the 50Hz

component of a power system
waveform corrupted by noise.

Where, N is the number of samples in the discrete rime sequence x(n), m is

called the harmonic index.
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 37
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 38
Use of Fourier transform
x(t) y(t)
LTIS, h(t)
Input Output

Frequency Response

Frequency reponse of a LTI System determines How the

frequency contents of input signal is passed to the out put.

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 39

 Example
x(t) y(t)
a) Find the impulse response of LTIS, h(t)
the system [ω= ωon]. Input Output

b) If x(t)=v(t)= 10(Sint-0.5
Sin2t+0.33 Sin3t) find out y(t)
ωo=1 and R/L=2
x(t) = v(t) = 10 sint -5 sin 2t +3,3 sin 3t
c) Draw the input v(t), amplitude and phase characteristic of
H(ω), and output signal of Vo(t).

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 40

Solution

x(t) y(t)
LTIS, h(t)
Input Output

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 41

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 42
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 43
 Input Signal Output Signal

90°
07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 44
 V(t)

R
S

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 45

 THANKS to
COMPUTER RELAYING FOR POWER Digital protection and signalling
SYSTEMS
MP Moore
Arun G. Phadke
James S. Thorp

07/04/2020 ELK508E,O USTA, ITU ELECTRICAL ENGINEERING 46