Page 1 of 5
European Journal of Business &
Social Sciences
Available at https://ejbss.org/
ISSN: 2235-767X
Volume 07 Issue 04
April 2019
Available online: https://ejbss.org/ P a g e | 2032
Estimation of Frequency in Power System Network
Using Linear and Nonlinear Techniques
RAMESH.D.
M.Tech – Power System Engineering,
PRIST Deemed University, Madurai, Tamilnadu
Abstract –
In an electrical power system frequency is an important
parameter. The frequency of operation is not constant but it
varies depending upon the load conditions. In the operating,
monitoring and controlling of electric device power system
parameters are having great contribution. So it is very important
to accurately measure this slowly varying frequency. Under
steady state conditions the total power generated by power
stations is equal to system load and losses. Frequency can
deviate from its nominal value due to sudden appearance of
generation-load mismatches. Frequency is a vital parameter
which influences different relay functionality of power system.
This study was made to estimate the frequency of measuring
voltage or current signal in presence of random noise and
distortion. Here we are first using linear techniques such as
complex least mean square (LMS), least square (LS) and
recursive least square (RLS) algorithm for measuring the
frequency from the distorted voltage signal. Then comparing
these results with nonlinear techniques such as nonlinear least
mean square (NLMS), nonlinear least square (NLS), nonlinear
recursive least square (NRLS) algorithms. The performances of
these
Algorithms are studied through simulation.
Key words: Frequency, linear techniques, Non linear techniques
I.INTRODUCTION
Frequency in electrical power system is an important
operating parameter which is required toremain constant because
it reflects the whole situation of the system. Frequency can show
the activeenergy balance between generating power and load.
Therefore frequency is considered as an indexfor operating
power systems in practical. In power system frequency is not
constant but changes according to load condition. Ideally this
frequency should be constant, but due to noise, sudden
appearance of load–generation mismatches and increasing use of
nonlinear load, the frequency of operating system is not
constant. Component reactance change results due to deviation
in system frequency from its desired value which influences
different relay functionality of power system such as server
damage or reactive power reduction occurs in system devices.
So the frequency plays an essential role in operating, monitoring
and controlling of any power system device. Basically
digitalized samples of current or voltage signal are used for
available frequency estimation techniques. Generally, the
voltage signal is used for frequency estimation because it is less
contort than the line current. Considering the purely sinusoidal
power system voltage signal, the frequency can be defined as
time
between two zero crossing. However in practice, the measured
signals are available in the distorted form and many techniques
are available for estimation of the frequency. Further, many
power electronic equipment and arc furnaces etc. generate
disturbances in the power systems. Therefore it is necessary for
utilities to develop a realistic approach for measurement of
frequency in presence of noises. Since frequency variation is a
dynamic phenomenon, the conventional phasor estimation
techniques such as Discrete Fourier Transform (DFT), Least
Square (LS) and Kalman Filtering may not be appropriate for
achieving accurate frequency estimation under dynamic
conditions. Zero crossing technique, discrete Fourier transform,
phase lock loop, least square error; orthogonal finite impulse
response filtering, Kalman filtering and iterative methods are
some techniques in this area. Soft computing technique such as
genetic algorithm and artificial neural network are also use for
power system frequency estimation. Also for protection and
control of the power system devices it is essential to have a
knowledge of frequency, therefore it is necessary that how we
are estimating the frequency without significant delay. It is
therefore absolutely necessary to develop a realistic approach for
frequency measurement in presence of disturbances. This paper
represented estimation of frequency from a distorted voltage
waveform. By considering different situation of power system
the distortion of the system signal is further enhanced. An
algorithm used here are complex least mean square (LMS) using
three phase voltage signal, LS estimator and RLS estimator. The
first two estimators use batch processing and third one is online
Page 2 of 5
European Journal of Business &
Social Sciences
Available at https://ejbss.org/
ISSN: 2235-767X
Volume 07 Issue 04
April 2019
Available online: https://ejbss.org/ P a g e | 2033
processing. After that frequency estimation is done using some
nonlinear techniques such as
Nonlinear Least Mean Square (NLMS), Nonlinear Least Square
(NLS), and Nonlinear Recursive Least Square (NRLS). Then by
comparing the results of both linear and nonlinear method.
II.MATHEMATICAL ANALYSIS-LINEAR ESTIMATION
Power system frequency estimation
In power system with no loss performance is considered
for estimation. So it is significantto have a purely sinusoidal
voltage or current signal for a better power quality. But in
practical, itdecline due to over voltage, variation in frequency,
source type, under voltage, harmonics, generation load
mismatches. Therefore it is needed to estimate the frequency fast
and accurately for better power quality in presence of noise and
other disturbances .Digitized samples of supply voltage are used
in most of the methods for frequency estimation in power
system. Basicallyfrequency is defined as the time between two
zero crossing of a purely sinusoidal voltage. But in real, signals
are not available in pure form but are distorted. Hence there are
various methods for estimation of frequency. Zero crossing
detection and calculation of the number of cycles within a
particular time interval are some basic methods for finding the
frequency of a purely sinusoidal voltage waveform. Discrete
Fourier transformation (DFT), Kalman filtering, least square
error method, and iterative approaches (2, 6-11) are some of
well-known approaches in this area. In this chapter complex
Least Mean Square, Least Square and Recursive Least Square
has been implemented for power system frequency estimation.
Frequency Estimation Using Least Square (Ls) Algorithm
The least square is a standard method for the
approximate solution of over determined systems. Over
determined system is the system which having more no. of
equations than unknowns. Least square means the overall
solution minimizes the sum of the squared error. Least square
approach has two types linear and nonlinear depending on
whether residuals are linear in all unknowns or not. Linear least
square is the linear combination of the parameters i.e. it consists
of the linear equations. Linear least square has a unique solution
Simulation result of Least Square (LS) Algorithm:
The figure shows the plot of frequency Vs No. of
iteration. Here estimated frequency comes to be
50 Hz.
Least square method is used here for frequency estimation. By
using Matlab simulation has been done. This method has good
accuracy but convergence time is more. There is delay of 2 to 3
cycles in frequency estimation.
Frequency Estimation using Recursive Least Square (RLS)
Algorithm
A distorted power system signal with noise can be
given as,
y(t) A1 sin(0t 1) (t)
Simulation result of Recursive Least Square (RLS) Algorithm:
The Below figure shows the frequency estimated Vs
No. of iteration. In recursive least square method frequency
estimated comes to be 50 Hz. This method has a higher
convergence rate than the previous one. Also time required to
estimate the frequency is less with good accuracy.
Frequency Estimation using Least Mean Square (LMS)
Algorithm
The least-mean-square (LMS) algorithm simplification
of the gradient vector and its computation is done by modifying
the objective function. Due to its computational simplicity The
LMS algorithm, is widely used in various applications of
adaptive filtering. In order to construct a range for the
convergence factor the convergence characteristics of the LMS
algorithm are studied that will guarantee stability .The
convergence speed of the Least Mean Square algorithm is
dependent on the Eigen value spread of the input signal
correlation matrix. The LMS algorithm is mostly used algorithm
in adaptive filtering for lots of reasons. The main features are
less complexity, unbiased convergence in the mean to the
There was a problem loading this page. Retrying...
Page 3 of 5
European Journal of Business &
Social Sciences
Available at https://ejbss.org/
ISSN: 2235-767X
Volume 07 Issue 04
April 2019
Available online: https://ejbss.org/ P a g e | 2034
Wiener solution, proof of convergence in stationary environment
and stable nature.
Steps for Frequency Estimation using Least Mean Square (LMS)
Algorithm
1. Initialize A , , , , , R , W , f ,
2. Generate power system signal
3. Using initial values of the weight vector estimate the
discredited signal
4. Error=Actual signal – Estimated signal
5. Update the step size and auto correlation matrix
6. Update weight vector
7. Go to step 4, until final iteration is reached
8. Estimate the fundamental frequency
Simulation result of Least Mean Square (LMS) Algorithm:
Simulation result of Least Mean Square (LMS) Algorithm:
LMS algorithm is applied to estimate the fundamental frequency
from the sampled value of the phase voltage signal. Parameters
of algorithm are 0.010.018, Pinitial , r=0.99, =0.01
And value changes according to sampling instant.
Figure shows the simulation result obtained by
implementing the above described LMS algorithmic Matlab.
Here in graph step change in frequency is given and
performance is observed. From the figure it is seen that during
step change in frequency from 50 Hz to 49 Hz there is delay of
0.001second. Accuracy and speed of estimation is satisfactory
even in presence of noise and harmonics. Advantages of this
algorithm are simplicity in formulation and computational
efficiency
IV.MATHEMATICAL ANALYSIS-NONLINEAR
ESTIMATION
Power system frequency estimation using Nonlinear Least
Square (NLS) Algorithm
Nonlinear Least Square estimation is the approach used
for the estimation of Fundamental frequency. Here frequency is
estimated by minimizing the squared error between actual signal
and assumed signal. For the model y(k) is the actual signal,(k)
is the system structure matrix and (k) is vector of parameters to
be estimated.
Simulation result of Nonlinear Least Square (NLS) Algorithm:
Accuracy of the algorithm decides the robustness of the
1-D search. The typical behavior of the norm of error as a
function of frequency is shown in the figure below. Here the
search is carried out by varying the fundamental frequency 0 f in
the steps of 0.1Hz .True frequency is the frequency at which the
minimum of the curve occurred. For the frequency estimation
we considered a voltage waveform having 10% total harmonics
and additive random noise which was given as
Here 3.2 KHz is the sampling frequency used for computer
simulation and the fundamental frequency is taken as 50Hz. In
power system half-wave symmetry waveforms are very common
soit does not contain even numbered harmonics and triplen
harmonics. Triplen harmonics are the 3rd and multiple of
fundamental harmonics. Using above sampling frequency,
maximum 32 harmonics can be taken into consideration without
any aliasing, which is suitable for the any application of power
system environment. Using above voltage signal we estimated
the frequency by nonlinear least square technique by varying the
frequency of voltage signal from 48.5 to 51.5Hz. As seen by
below figure simulation result we found that minimum error
occurred at 50.1Hz which is the estimated frequency.
There was a problem loading this page. Retrying...
