G04123844

International Journal of Engineering Inventions
e-ISSN: 2278-7461, p-ISSN: 2319-6491
Volume 4, Issue 12 [August 2015] PP: 38-44
www.ijeijournal.com Page | 38
Single Phase Inverter with Selective Harmonics Elimination
PWM Based on Secant Method
Essam Hendawi1, 2
1
Department of Electrical Engineering, Faculty of Engineering, Taif University, Kingdom of Saudi Arabia
2
Electronics Research Institute, Egypt
Abstract: Selective harmonics can be eliminated by controlling the inverter switches at optimized switching
angles in single phase dc-ac inverters. These switching angles are estimated by solving a set of nonlinear
equations. This paper presents a novel technique for solving these equations based on secant method. The
proposed algorithm simplifies the numerical solution of the nonlinear equations, solves them fast and achieves
convergence of solution at different starting vector of angles. The proposed method is applied with different
modulation index and different number of equations, i.e. different number of switching angles to achieve the
elimination of certain harmonics. High performance of the inverter is validated through simulation results.
Keywords:Selective harmonics elimination, Secant method, Switching angles.
I. INTRODUCTION
Single phase dc-ac inverters are widely used in many applications such as AC motor drives, power
supplies and uninterruptible power supplies [1]. Half bridge and full bridge topologies can be found but full
bridge topology are more common. The output voltage of the inverter is required to be as close as possible to
sinusoidal shape. Conventional sinusoidal pulse width modulation SPWM can be applied but selective
harmonics elimination pulse width modulation (SHEPWM) technique is more advantageous than SPWM as
listed in the literature [1-5].In SHEPWM, number of low order harmonics is selected to be eliminated based on
the requirements of application. Then a set of nonlinear equations called “transcendental equations” are solved
numerically to get the switching angles. These switching angles are applied to generate the control signals of
the inverter switches. Equal area algorithm, Newton-Raphson algorithm, pattern search method and genetic
algorithm can be applied to solve these equations [2, 3, 6]. The main challenges are the starting vector of
solution and the convergence of the solution. This paper proposes a novel technique for solving the
transcendental equations based on secant method. The proposed algorithm simplifies the numerical solution of
the nonlinear equations, solves them fast and achieves convergence of solution at different starting vector of
angles. The switching angles are estimated using the proposed method. The switching angles are thenutilized to
generate the control signals which are applied to unipolar single phase full bridge inverter. High performance of
the inverter is validated through simulation results.
II. PRINCIPLE OF SHEPWM INVERTERS
Schematic of full bridge single phase inverter is shown in Fig. 1.The states of the four switches and the
corresponding load voltage are summarizedin Table 1.Two techniques for selective harmonics elimination
Fig. 1 Single phase inverter

Single Phase Inverter with Selective Harmonics Elimination PWM Based on Secant Method
Table 1 States of the inverter switches
Switches state Vo
T1 and T2 are “on” - T3 and T4 are “off” Vdc
T3 and T4 are “on” - T1 and T2 are “off” -Vdc
T1 and T3 are “on” - T2 and T4 are “off” 0
T2 and T4 are “on” - T1 and T3 are “off” 0
PWM can be applied; bipolar PWM technique and unipolar PWM technique [1]. In the bipolar PWM
technique, the first two states in Table 1 are only utilized. The four states are applied in unipolar PWM
technique which is more effective than bipolar PWM technique when harmonics elimination is required.
Therefore unipolar PWM technique is applied in this paper. The output voltage of the unipolar PWM inverter is
shown in Fig. 2 with N switching angles (ɵ1 , ɵ1 , ɵ1 … ɵN) , (N-1) harmonics can be eliminated. The output
voltage can be represented in Fourier series as a function of the switching angles as
𝑣0 𝜔𝑡 =
4𝑉𝑠
𝜋
(
sin 𝑘𝜔𝑡
𝑘
[cos 𝑘𝜃1 −
∞
𝑘=1,3,5
cos 𝑘𝜃2 + cos 𝑘𝜃3 − ⋯ . . ] ) (1)
Only odd harmonics exist while even harmonics do not exist since the output voltage is even function in its
nature. The conditions for fully elimination of certain number (N-1) of odd harmonics are:
(−1)𝑖+1
cos 𝜃𝑖 −
𝜋
4
𝑀𝑎 = 0
𝑁
𝑖=1
(2)
(−1) 𝑘+1
cos 𝑛𝜃𝑖 = 0
𝑁
𝑘=1
(3)
Where Ma is the modulation index which is defined as
Ma = V1 / Vdc , n = 3, 5, 7, ….., 2N – 1 and θ1< θ2< θ3 …. <θN< 90o
For instance if it is desired the 3rd
,5th
,7th
,9th
, 11th
and 13th
harmonics, equations (2) and (3) can be written as:
(−1)𝑖+1
cos 𝜃𝑖 −
𝜋
4
𝑀𝑎 = 0
7
𝑖=1
(4)
(−1) 𝑘+1
cos 𝑛𝜃𝑖 = 0
7
𝑘=1
(5)
Where 'n' takes the values 3, 5, 7, 9, 11, 13. These will lead to the following seven equations which should be
solved simultaneously:
cos θ1 − cos θ2 + cos θ3 − cos θ4 +cos θ5 - cos θ6 + cos θ7 −
π
4
M
a
= 0 (6-a)
cos 3θ1 − cos 3θ2 + cos 3θ3 − cos 3θ4 + cos 3θ5 - cos 3θ6 + cos 3θ7 =0 (6-b)
cos 5θ1 − cos 5θ2 + cos 5θ3 − cos 5θ4 + cos 5θ5 - cos 5θ6 + cos 5θ7 =0 (6-c)
cos 7θ1 − cos 7θ2 + cos 7θ3 − cos 7θ4 + cos 7θ5 - cos 7θ6 + cos 7θ7 =0 (6-d)
cos 9θ1 − cos 9θ2 + cos 9θ3 − cos 9θ4 + cos 9θ5 - cos 9θ6 + cos 9θ7 =0 (6-e)
cos 11θ1 − cos 11θ2 + cos 11θ3 − cos 11θ4 + cos 11θ5 - cos 11θ6 + cos 11θ7 =0 (6-e)
cos 13θ1 − cos 13θ2 + cos 13θ3 − cos 13θ4 + cos 13θ5 - cos 13θ6 + cos 13θ7 =0 (6-f)
These nonlinear transcendental equations are solved numerically using the proposed algorithm by applying
secant method.
Fig. 2 Output voltage of single phase full bridge inverter with unipolar PWM
0.06 0.062 0.064 0.066 0.068 0.07 0.072 0.074 0.076 0.078 0.08
-80
-60
-40
-20
0
20
40
60
80
Loadvoltage(V)

III. PROPOSED ALGORITHM USING SECANT METHOD
As a brief description of Newton-Raphson method to solve systems of non-linear equations, it is required to
determine a Jacobian matrix “J” which is defined for N equations as follows:
J =
𝜕𝑓1
𝜕𝜃1
𝜕𝑓1
𝜕𝜃2
𝜕𝑓1
𝜕𝜃3
⋮
𝜕 𝑓1
𝜕 𝜃 𝑁
𝜕𝑓2
𝜕𝜃1
𝜕𝑓2
𝜕𝜃2
𝜕𝑓2
𝜕𝜃3
⋮
𝜕 𝑓2
𝜕 𝜃 𝑁
⋮ ⋮ ⋮ ⋮ ⋮
𝜕𝑓 𝑁
𝜕𝜃1
𝜕𝑓 𝑁
𝜕 𝜃2
𝜕𝑓 𝑁
𝜕𝜃3
⋮
𝜕 𝑓 𝑁
𝜕 𝜃 𝑁
(7)
This Jacobian matrix is very complex since it needs a large number of partial differentiations especially when
several number of harmonics are desired to be eliminated i.e. high value of “N”. Therefore very complex
calculations will be needed and divergence of solution may occur. With the proposed method, each partial
differentiation is replaced by the following equation:
𝜕𝑓 𝑖
𝜕𝜃 𝑘
≈
𝑓 𝑖 𝜃1,𝜃2,….𝜃 𝑘+ ∆𝜃,…..,𝜃 𝑁
∆𝜃
− 𝑓 𝑖 𝜃1,𝜃2,….,𝜃 𝑘,…..,𝜃 𝑁
∆𝜃
(8)
i and k take the values 1, 2, …., N ∂f2/ ∂θ3represents element J23in the Jacobean matrix
𝜕𝑓2
𝜕𝜃3
≈
𝑓2 𝜃1, 𝜃2, … . 𝜃3 + ∆𝜃, … . . , 𝜃 𝑁
∆𝜃
− 𝑓𝑖 𝜃1, 𝜃2, … ., 𝜃3, … . . , 𝜃 𝑁
∆𝜃
Where “Δθ” is a small increment. The unknowns θ1, θ2 , … θN are assigned initial values as:
θ0 = [θ1θ2 θ3 …. θN]T
Flow chart that summarizes the proposed algorithm is given in Fig. 3. Equation (8) is applied to determine each
element of the Jacobian matrix, then the elements are combined together to estimate the Jacobian matrix “J” as
mentioned in (7). An update of the vector θ is carried out using the relation:
𝜽 𝒏+𝟏 = 𝜽 𝟎 + 𝑱−1
. 𝒇 𝜃𝑛
The algorithm utilizes θ1as an initial vector for the next cycle which will repeat until the stopping criteria is
achieved. this criteria is
max 𝑓𝑘 (𝜽) ≤ 𝜀1 , 𝑘 = 1, 2, … , 𝑁,
max (𝜃 𝑘) 𝑛+1 − (𝜃 𝑘) 𝑛 ≤ 𝜀2 , 𝑘 = 1, 2, … , 𝑁 ,
Or the maximum number of iterations is exceeded which means divergence of solution and another initial
solution vector θ0 should be selected.

Fig. 3 Flow chart of the proposed algorithm
IV. SIMULATION RESULTS
The proposed method estimates the switching angles θ1 θ2 θ3 …. θN . These switching angles are then
utilized in an m-file in Matlab software program to generate the control signals for the inverter switches. Two
Matlab m-files are built to carry out the algorithm at different number of harmonics that are desired to be
eliminated. The first file is for constant data including dc voltage, desired output frequency and initial vectors
of solution. The second file is for the proposed described algorithm. The dc source is 60 V and the load is R-L
where R = 10 Ω and L = 40 mH. The increment “Δθ” is chosen as 0.05 rad The control signals are adjusted
so that the load frequency is 50 Hz. Different cases are included from N = 3 to N = 11. The switching angles
are given in Table 2. Selected cases are studied; N=7, N=10 and N=11. Load voltage and current for N=7,
N=10 and N=11 are shown in Figures 4-6 respectively. The modulation index “Ma” is adjusted at 0.85 The
associated spectrum analysis of load voltage is presented in Figures 7-9 respectively.It is obvious the harmonics
of order 6, 9 and 10 are eliminated for N=7, N=10 and N=11 respectively. The load current is very close to
sinusoidal shape as “N” increases. To expect the switching angles using the proposed method and reduce the
No
No
Yes
Yes
Update the solution vector θ
Stop and use the final vector
θ to generate the control
signals
Is the No. of iteration
exceeded?
Is the conversion
met?
Select another initial
vector of solution θo
Use the new solution
vector θ as an initial
vector
Determine the Jacobean matrix J
using equation (8)
Set initial vector of solution θo and
constant data

iteration cycles especially with large values of “N”, variation of switching angles with “N” is given in Fig. 10.
Therefore the initial values of the switching angles can be selected properly
Fig. 4 Load voltage and current, N = 7, Ma = 0.85
0.06 0.07 0.08 0.09 0.1 0.11 0.12
-80
-60
-40
-20
0
20
40
60
80
Loadvoltage(V)
0.06 0.07 0.08 0.09 0.1 0.11 0.12
-6
-4
-2
0
2
4
6
time (sec)
Loadcurrent(A)
0.06 0.07 0.08 0.09 0.1 0.11 0.12
-80
-60
-40
-20
0
20
40
60
80
Loadvoltage(V)
0.06 0.07 0.08 0.09 0.1 0.11 0.12
-6
-4
-2
0
2
4
6
time (sec)
Loadcurrent(A)
0.06 0.07 0.08 0.09 0.1 0.11 0.12
-80
-60
-40
-20
0
20
40
60
80
Loadvoltage(V)
0.06 0.07 0.08 0.09 0.1 0.11 0.12
-6
-4
-2
0
2
4
6
time (sec)
Loadcurrent(A)

Fig. 7 spectrum analysis of load voltage, N = 7, Ma = 0.85
0 10 20 30 40 50 60 70 80 90 100
0
5
10
15
20
25
30
35
40
45
50
Order of Harmonic
Magnitudebasedon"BasePeak"-Parameter
0 10 20 30 40 50 60 70 80 90 100
0
10
20
30
40
50
Order of Harmonic
0 10 20 30 40 50 60 70 80 90 100
0
5
10
15
20
25
30
35
40
45
50
Order of Harmonic

Fig. 10 variation of switching angles versus “N”
V. CONCLUSION
Selective harmonics elimination is achieved by adjusting the switching angles of the inverter. The
switching angles have been estimated using the proposed algorithm which is based on secant method. The
switching angles are utilized to generate the control signals of the inverter switches. Simulation results verify
the effectiveness of the proposed algorithm. Selective harmonics in the inverter output are eliminated and the
load is very close to sinusoidal shape. The proposed algorithm can be applied for any number of harmonics.
A guide of the initial vector of switching angles is proposed to minimize the iterations especially with large
number of harmonics.
REFERENCES
[1] Muhammad H. Rashid, Power Electronics Handbook”, 2nd
Edition, 2007,Elsevier.
[2] K. Keerthivasan, M. Narayanan, V. SharmilaDeve and J. Jerome, “Estimation of Optimal Single phase
Inverter’s Switching angles Using Pattern Search Method”, International Conference on Computer
Communication and Informatics (ICCCI -2012), Jan. 10 – 12, 2012, Coimbatore, India.
[3] Tamer H. Abdelhamid and Khaled El-Naggar, “Optimum Estimation of Single Phase Inverter’s
Switching Angles Using Genetic Based Algorithm,” Alexandria Engineering Journal, 2005, Vol. 44,
No.5, pp 751-759.
[4] J. E. Denis and H. Wolkowicz, “Least change secant methods, sizing, and shifting,” SIAM J. Numer.,
1993. Anal., vol. 30, pp. 1291–1314
[5] Muhammad H. Rashid "Power Electronics (circuits, devices and applications)", 1988, 1st
Edition,Prentice-Hall.
[6] J. W. Chen and T. J. Liang, “A Novel Algorithm in Solving Nonlinear Equations for Programmed PWM
Inverter to Eliminate Harmonics”, 1997, 23rd
International Conference on Industrial Electronics, Control
and Instrumentation conference IECON.
3 4 5 6 7 8 9 10 11
10
20
30
40
50
60
70
80
90
N
Switchingangles

G04123844

Related slideshows

More Related Content

G04123844