IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 21, NO. 3, JULY 2006 1743
A Simple Energy Operator Computational
Method for Voltage Flicker Assessment
Mohamed Amin Eldery, Student Member, IEEE, Ehab F. El-Saadany, Senior Member, IEEE, and
Magdy M. A. Salama, Fellow, IEEE
Abstract—In this paper, a novel algorithm, based on calculating
the energy operator of a sinusoidal waveform to track the voltage
flicker, is presented. The mathematical derivation of the proposed
algorithm different circuit designs required to realize it are de-
scribed. The proposed algorithm is fast and robust and requires
only a few samples to calculate the energy. The simulation and the
experimental results prove that the new algorithm can successfully
track the envelope. There are no spikes in the envelope even when
the voltage signal is suddenly changed at its peak value.
Index Terms—Digital signal processor (DSP), envelope tracking,
power quality monitoring, sag, swell, teager energy operator,
voltage flicker.
I. INTRODUCTION
M
ODERN end-use equipment is very sensitive to voltage
fluctuation and flicker. In many cases, mitigating the
voltage flicker helps to prevent equipment malfunction. Flicker
mitigation techniques depend on injecting a certain amount of
reactive power, defined by the difference between, the reference
value and the measured voltage. As a result, the measurement
accuracy is crucial. Voltage flicker measurement depends prin-
cipally on the accuracy of the envelope tracking algorithm. The
envelope tracker should be accurate, robust, and fast with the
least mathematical burden.
The voltage envelope is either calculated or estimated. Dif-
ferent signal processing algorithms, such as, the Fast Fourier
Transform (FFT) [1], [2] and the Wavelet [3], [4] are adopted to
calculate the voltage envelope. However, these algorithms intro-
duce a lag in the envelope tracking that is equal to the length of
the used window. It is this delay that imposes some limitations
on the on-line application of these algorithms.
The envelope of the voltage signal can be estimated by var-
ious estimation methods such as the Kalman Filter (KF) [5],
the Least Absolute Value (LAV) [6], the Simulated Annealing
(SA) [7], and the ADAptive LInear NEuron (ADALINE) [8].
Although the KF is fairly accurate, it has a high mathematical
burden which limits its use for on-line tracking. The LAV and
SA algorithms require that the flicker waveform is known in ad-
vance which is not a realistic assumption. Although the ADA-
LINE is efficient and has a fast convergence, compared to other
estimation algorithms, the ADALINE is still considered mathe-
matically cumbersome.
Manuscript received December 27, 2004; revised December 24, 2005. Paper
no. TPWRD-00621-2004.
The authors are with the Department of Electrical and Computer Engi-
neering, University of Waterloo, Waterloo, ON N2L 3G1, Canada (e-mail:
maaelder@uwaterloo.ca).
Digital Object Identifier 10.1109/TPWRD.2006.874106
In [9], it is proven that the energy contained in a sinusoidal
signal is proportional to the square of the signal amplitude.
Thus, it is proposed in [10] to use the energy operator to
track the envelope by tracking the amplitude of the signal.
The Teager Energy Operator (TEO) is successfully applied
to achieve this task since the TEO calculates the energy, and
in turn, the amplitude, by only three samples [11]. Since,
the algorithm does not depend on any kind of estimation or
optimization, the algorithm is considered suitable for on-line
tracking. Moreover, the delay between the actual envelope and
the tracked envelope is within two samples only. However, the
algorithm’s practical application indicates that large spikes are
embedded in the tracked envelope which is a fatal drawback in
using these measurements for flicker mitigation techniques. In
addition, these spikes reach very high values when the level of
the amplitude is suddenly changed [10], [12].
In this paper, tracking the fluctuation in the voltage waveform
is achieved by a novel algorithm to calculate the energy oper-
ator. The proposed algorithm attains the same TEO advantages:
simple, fast and independent of estimation techniques. Yet, the
new algorithm overcomes TEO drawbacks, insensitive to noise
and immune to severe spikes in voltage magnitude.
This paper is organized as follows: Section II describes the
mathematical modeling of flicker. Section III presents the TEO
along with the mathematical derivation of the proposed algo-
rithm. Different circuit designs of the proposed algorithm are
discussed in Section IV. In Section V, the proposed algorithm
is used to track the envelope of various signals modulated by
different waveforms. Section VI shows the experimental verifi-
cation of the proposed algorithm. Finally, Section VII concludes
the paper.
II. VOLTAGE FLICKER GENERATION
Voltage flicker is characterized by slow changes in the root
mean square (rms) of the voltage that cause the light flickering.
Voltage flicker can be represented by either a modulated signal
or generated experimentally by utilizing nonlinear loads.
A. Flicker Simulation
The slow changes in the are represented by the following
amplitude modulation:
(1)
where is the amplitude of the fundamental, is the funda-
mental frequency and is the modulating signal that can have
different shapes such as square wave, sinusoidal wave, triangle
0885-8977/$20.00 © 2006 IEEE