Identifying Harmonics by Empirical ModeDecomposition for Effective Control of ActiveFilters During Electric Vehicle Charging
B. V. Dhananjay and T. Senthil
B.V. Dhananjay,
Research Scholar,Vinayaka Mission’s University,Salem,India
Dr.T.Senthilkumar
,
Professor, Automobile Engineering,Bharathidhasan University,Trichirapalli, India
Abstract-
This paper provides Hilbert HuangTransform(HHT) method (an empirical modedecomposition(EMD)) for identifying the presence of harmonics during electric vehicle battery chargingwhen harmonics are generated into the electric line,due to switching actions of the power electronics.Activation of the active filters based on the differencebetween load current and fundamental currentmeasured from the line is done. By using active powerfilter (APF) injection of the required current tominimize the harmonics is done. As part of simulation, the accuracy of the HHT is above 95%.By correctly recognizing the harmonics using HHTand injecting the compensating current into the line,the charging time of the battery can be reduced. Thereduction in the charging time also depends on thebattery condition.
Keywords-Hilbert Huang Transform; active power filter.
I
INTRODUCTIONThe battery is the primary source of electricalenergy. It stores chemicals. Two different types of lead in an acid mixture react to produce anelectrical pressure. This electrochemical reactionchanges chemical energy into electrical energy. A battery can be of primary cell, secondary cell, wetcharged, dry charged and low maintenance type. Afully charged battery contains a negative plate of sponge lead(Pb), a positive plate of leaddioxide(Pbo
2
) and an electrolyte of sulphuric acid(H
2
So
4
) and water (H
2
o). During charging, sulphateleaves the plates and combines with hydrogen(H
2
)to become sulphuric acid (H
2
So
4
). Free oxygencombines with lead on the positive plate to formlead dioxide. Gassing occurs as the battery nearsfull charge and hydrogen bubbles out at thenegative plates, oxygen at the positive. Factorsaffecting charging are temperature, state of charge, plate area, impurities, gassing.Electric vehicles (EV) will become anattractive alternative to internal combustion enginevehicles in the event that their range can beextended. One way to achieve this in the short termis to provide a fast charger infrastructure. Such astructure would provide greater mobility for the EVuser, since during short stops (<1 hour) the EV batteries could be charged from typically 20 to 80% of nominal charge. This would significantlyextend the EV range.Fast charger infrastructurecost is high. Chargers adversely affect the grid power quality due to presence of power electronicloads like diode rectifiers and thyristor bridgeconverters in the distribution network that result involtage distortion and current harmonics, (Akagi1996).High increase of problems in the electric power distribution networks due to the presence of harmonics. Loads that use switching control withsemiconductor devices are the main cause. One of the most important tools for correcting the lack of electric power quality are the active power filters(APF), (Udom et al. 2008). The objective of thiswork has been proving that back propagationneural networks, previously trained with a certainnumber of distorted waveforms, are an alternativeto the rest of the techniques used and proposed atthe present time for controlling the APF's, as theones based on the use of the Fast Fourier Transform (FFT). A large number of these controltechniques are based on ANN’s, (Pecharanin et al.1994).II
MATERIALS AND METHODSA
Materials
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 9, No. 9, September 2011109http://sites.google.com/site/ijcsis/ISSN 1947-5500
Figure 1shows a three-phase diagram of an HHT controlled shunt APF. A load currentsignal
i
L
is acquiredand used by the ANN to obtainthe distortion current waveform as reference signalfor the control of the APF.The power converter injects the necessary compensation current
i
L
in the power circuit, achieving thus a sinusoidal sourcecurrent.
Figure 1
APF control using HHT
B
Methods
Empirical Mode Decomposition (Huang) andHilbert Transform
A signal can be analyzed in details for itsfrequency, amplitude and phase contents by usingEMD followed by HT (Jayasree et al. 2010 andStuti et al. 2009), The EMD produces the monocomponents called IMFs from the original signal.In a given frame of signal, there can be manyIMFs. Each IMF will contain a wave form of different amplitude. Hilbert Transform is appliedon an IMF to obtain, IF and IA. It is mandatory thata signal be symmetric regarding the local zeromean, and should contain same number of extremeand zero crossings.The steps involved in EMD of a signal X(t) withharmonics into a set of IMFs are as follows.1.
Identify all local maxima of X(t). Connect the points using a cubic spline. The interpolatedcurve obtained. The upper line is called theupper envelope (Maximum_envelope).2.
Identify all local minima of X(t) connect the point using a cubic spline.. The lower line iscalled the lower envelope(Minimum_envelope) obtained by cubic spline.
3.
Compute the average by:
2 b)(aM
+=
(1)Where a = Maximum_envelope and b =Minimum_envelope.4.
Obtain a new signal using the followingequation:
(t)MX(t)(t)h
1111
−=
(2)Where h
11
(t) is called first IMF. SubsequentIMF’s had to be found if there are someovershoots and undershoots in the IMF. Hence,the envelope mean differs from the true localmean and h
11
(t) becomes asymmetric.In order to find the additional IMF’s, h
11
(t)is taken as the new signal. After n
th
iteration, wehave:
(t)M(t)h(t)h
1n1)1(n1n
−=
−
(3)Where M
1n
(t) is the mean envelop after the n
th
iteration and h
1(n-1)
(t) is the difference betweenthe signal and the mean envelope at the (k-1)
th
iteration.5.
Calculate C2F as follows:
n1
IMFC2F
=
(4)Where
n
IMF
= final IMF obtained
1)(nn2
IMFIMFC2F
−
+=
(5)Similarly,
11)(nnn
IMF.......IMFIMFC2F
+++=
−
(6)Where C2F
n
is the original signal.6.
Calculate F2C as follows:
11
IMFF2C
=
(7)
212
IMFIMFF2C
+=
(8)
n21n
IMF.......IMFIMFF2C
+++=
(9)Where F2C
n
is the original signal.7.
Hilbert transform is applied for eachIMF and analytical signal is obtained.A complex signal is obtained from eachIMF:
imag(IMF)real(IMF)MF)Analytic(I
+=
(10)
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 9, No. 9, September 2011110http://sites.google.com/site/ijcsis/ISSN 1947-5500
8.
Instantaneous frequencies are obtainedfrom analytical signal using
π
2
π
)1)))conj(X(t1)X(t(angle(0.5
IF
×+−×+−×
=
(11)9.
Instantaneous amplitudes are obtainedfrom the analytical signal using thefollowing
22
imag(IMF)real(IMF)IA
+=
(12)
III
EXPERIMENTAL SIMULATION
Figure 2
Power circuit with HHT controllingactive power filter
The model of Figure2 has been createdusing Matlab 10. Different sets of parameters have been employed at the power circuit and APF. Inmost cases the reference current obtained by theHHT controller was accurate enough to enable theAPF to compensate harmonic distortion. If anelevated content of high order harmonics were present in the load current, the HHT controller helps in obtaining reference signal.
System parameter used:Power circuit:
Phase voltage = 220 VRMS ,Frequency = 50 Hz , Source resistance = 0.1
Ω
,Load resistance = 20
Ω
, Load inductance = 30 mH
APF:
Vdc = 500 V, R = 10
Ω
, L = 30 mH,Switching frequency = 40 KHz.IV
RESULTS AND DISCUSSIONMore than 500 different harmonicswaveforms (Figure 3) have been used in HHTanalysis with different load changes.The results of a simulation are presented where a step load changeoccurs at time 60 ms. One additional resistance isconnected in parallel with the load, increasing thetotal load current.Figure 4 shows the EMD process. In thesample harmonics signal considered, only oneinstantaneous mode function is present. A flatresidue signal is also presented. This plot is onlyfor 1000 samples. This will be repeated for theremaining length of the signal. Figure 5 shows theextraction of different signals present from the finelevel to coarse level. Similarly, Figure 5 shows theextraction of different signals present from thecoarse level to fine level. Figure 7 presents theinstantaneous frequency present in every sample of the signal. Figure 8 presents the instantaneousamplitude present in every sample of the signal.Figure 9 and Figure 10 presents statistical values of instantaneous frequencies and instantaneousamplitudes. Based on the statistical values, theamount of harmonics will be estimated andappropriately, required compensating current will be injected into the line.V
CONCLUSIONA Hilbert Huang Transform method has been used at the control of a shunt active power filter. Based on the amount of harmonicsrecognition, the APF is activated. By correctlyinjecting the compensating current into the line, thecharging time of the battery can be reduced. Thecircuit has to be verified with the implementationof HHT in real time for improved charging of theEV battery. The reduction in the charging time alsodepends on the battery condition.
Figure 3
Sample plot of signals for harmonics
00.510246-400-2000200400
Time(Sec)Signal patterns
V o l t a g e ( V )
v=0.1pu,cy=50v=0.1pu,cy=50v=0.1pu,cy=50v=0.1pu,cy=50v=0.1pu,cy=50
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 9, No. 9, September 2011111http://sites.google.com/site/ijcsis/ISSN 1947-5500
Search History:
Result 00 of 00
00 results for result for
p.
Identifying Harmonics By Empirical Mode Decomposition For Effective Control Of Active Filters During Electric Vehicle Charging
This paper provides Hilbert Huang Transform (HHT) method (an empirical mode decomposition(EMD)) for identifying the presence of harmonics during electric vehicle battery charging when harmonics are generated into the electric line, due to switching actions of…
(More)
206
Reads
0
Readcasts
0
Embed Views
More From This User
Notes
Load more
