Real-Time Optimization of Thermal Cycling
Capability of Rotor Side Converter in Doubly Fed
Induction Generator- Based Wind Energy
Conversion System
Lina Alhmoud
Bingsen Wang
Department of Electrical and Computer Engineering
Michigan State University
2120 Engineering Building
East Lansing, MI 48824, USA
Emails: hmoudlin@egr.msu.edu; bingsen@egr.msu.edu
Abstract—The power command in wind energy conversion
system is subjected to the wind speed fluctuation, these power
fluctuation may cause significant thermal cycling of the (IGBT)
power converter with reduction in lifetime as a result. In this
work the (DFIG) is used as a starting point, average model is
adopted to replace the switching circuits in three phase back to
back PWM to make the simulation environment run faster and
a detailed thermal model for the converter is implemented and
simulated based on the state of the art on (IGBT) reliability,
the lifetime is estimated. This paper demonstrates that using low
pass filter (LPF) on power command can reduce the number of
thermocycles and improve the system lifetime.
NOMENCLATURE
Vs , Vr
RMS stator and rotor voltages.
is , ir
RMS stator and rotor currents.
s
Slip.
λs , λr
Stator and rotor flux linkages.
m1 , m2
Stator and rotor converter modulation.
va , vb , vc
3-phase supply voltage.
vd , vq , vα , vβ 2-axis supply voltages.
ωe , ωr , ωslip Supply, rotor, slip angular frequency.
P, Q
Active and reactive power.
θe , θs
tion.
Supply voltage, stator flux vector posi-
C
DC link capacitor.
E
DC link voltage.
Ls , Lr
inductance.
Stator and rotor per phase winding
Lls , Llr
Stator and rotor per phase winding
leakage inductance.
Lm
Machine magnetizing inductance.
Rs , Rr
resistance.
Stator and rotor per phase winding
L, R
side.
Inductance and resistance of supply
p
Number of machine pairs of poles.
J
inertia.
Te
electromagnetic torque.
Suffices, Superscripts
s, r
stator, rotor.
d, q
d-q reference frame.
a, b, c
Three phase reference.
∗
demand (reference) value.
I.
I NTRODUCTION
Reliability is the probability that a component will satisfactory perform its intended function under given operating
conditions, the average time of satisfactorily operation of a
system is called the mean time between failures (MTBF) and
the higher value of (MTBF) refers to a system of higher
reliability and verse a versa. Nowadays reliability is more of
concern than in the past especially for offshore wind turbines
since the access to offshore wind turbines in case of failures
is both costly and difficult. Power semiconductor devices are
often ranked as the most fragile components in a power
conversion system. The lifetime prediction of power (IGBT)
modules based on realistic mission is an important issue.
However, lifetime modeling of future large wind turbines is
needed in order to make reliability predictions about these new
wind turbines early in the design phase. By doing reliability
prediction in the design phase the manufacture can ensure that
the new wind turbines will live long enough.
This work presents reliability analysis of power electronic
converters for wind energy generation systems based on semiconductor power losses and aims to maximize semiconductor
lifetime using low pass filter (LPF), because (MTTF) will be
higher than without filter. We ought to reduce the number of
thermo cycles. To the best of our knowledge, no research has
been performed so far, has used real time control to improve
reliability using LPF. The key element in a power conversion
system is the power semiconductor devices that operates as
power switch, the improvement in power semiconductor device
is driving force behind the improved performance, efficiency,
size and weight of power conversion systems. As the power
density and switching frequency increase, thermal analysis
of power electronic system becomes imperative, the analysis
provides information on semi conductor rating, reliability,
lifetime calculation [1] [2]. A comprehensive thermal model
for power (IGBT) is developed into main three steps [3]:
First, the losses are calculated [4], the junction temperature
is evaluated, and the lifetime is estimated. The power loss
model, which is based on look up table for calculating the
switching and conduction losses are successfully built to
perform the thermal analysis. The parameters of the thermal
network are extracted from ABB HiPack IGBT Module 5SNE
0800M170100 characteristic data sheets. The exact modeling
approach describes a converters as a time varying system is
typically not applicable for control design purpose because
they are difficult to analyse and impractical for the simulation
of the relatively large systems, because they require simulation time steps much smaller than switching period, and a
result in a very computationally expensive simulation. thus,
an averaged model is suitable for electro thermal simulation,
it can be used to calculate the semiconductor losses at any
output current waveform, it considered faster and it can be
parameterized with conventional data sheet information. For
calculating the instantaneous temperature of the semiconductor
junctions at different load conditions a thermal model of the
inverter is necessary. A straightforward approach is to use a
network of thermal capacitance Cth and Rth , such networks
are easily applicable in any programming. Different cumulative
damage theories have been proposed for the purposes of
assessing fatigue damage caused by operation at any given
stress level and the addition of damage increments to properly
predict failure under conditions of spectrum loading. Collins,
in (1981), provides a comprehensive review of the models
that have been proposed to predict fatigue life in components
subject to variable amplitude stress using constant amplitude
data to define fatigue strength. The original model, a linear
damage rule, originally suggested by Palmgren (1924) and
later developed by Miner (1954) [5]. The linear theory, which
is still widely used, is referred to as Palmgren-Miner rule or
the linear damage rule. Estimates may be made by employing
Palmgren Miner rule along with a cycle counting procedure.
The rest of the paper is organized in four sections. Section
II discuss concepts, mathematical equations, implementation of
wind turbine characteristics, double Fed Induction Generator
(DFIG) and the average model for the converter, briefly,
the physical system is presented in section II, since losses
and junction temperature and its variation directly affect the
lifetime of the converter, furthermore, lifetime calculations
and reliability estimation are presented in section III, issues
related an adequate control for smoothing the power command
injected in (RSC) of (DFIG) is introduced in section IV, this
paper is finished by a conclusion and a list of references.
II.
I NTRODUCTION TO P HYSICAL SYSTEM M ODELING
A. Wind Turbine Characteristics
Wind turbines capture power from the wind by means of
aerodynamically designed blades and convert it to rotating
mechanical power. The output power Pm is dependent on the
power coefficient Cp . It is given by [6]:
Pm =
1 2 3
ρR υ Cp (λ, β)
2
(1)
and the tip speed ratio is defined as:
λ=
ωt R
υ
(2)
where ρ is specific air density (kg/m2 ), R is radius of the
turbine blade (m) , υ is the wind speed (m/s). ωt is turbine
speed , Cp is the coefficient of power conversion and β is the
pitch angle. The relation of the power coefficient Cp (λ, β) is
further expressed as [6]:
Cp (λ, β) = c1
−c5
c2
− c3 β − c4 e λ1 + c6 λ
λ1
(3)
where the coefficients c1 through c6 depend on the shape of
the blades and its aerodynamic performance of wind turbine,
and λ1 is defined as:
0.035
1
1
− 3
=
λ1
λ + 0.08β
β +1
(4)
B. Doubly-Fed Induction Generator
The Doubly Fed Induction Generator (DFIG) is considered
as one of the most popular topologies applied in wind power
systems, because of its very good characteristics. Its main
advantage is to adjust the speed of large system with power
converters of a third of full rating, this is because its rotor speed
converter (RSC) operates under slip frequency and it needs
only to support slip power to the overall system. However,
this slip frequency is much lower than the grid frequency
and insulated gate bipolar transistors (IGBTs) in the (RSC)
and grid side control (GSC) are susceptible to power cycling
failures. (DFIG) configuration is the best suited since it can
be controlled from rotor side as well as stator side, this is
possible since rotor circuit is capable of bidirectional power
flow as shown in Fig. 1. The following equations describe the
general model of (DFIG) [9]:
dλds
− ωλqs
dt
dλqs
vqs = rs iqs +
+ ωλds
dt
dλdr
= rr idr +
− (ω − ωr )λqr
dt
dλqr
= rr iqr +
+ (ω − ωr )λdr
dt
vds = rs ids +
vdr
vqr
(5)
(6)
(7)
(8)
C. Averaged Model of (PWM)
converter
The averaged model of three-phase back-to-back (PWM)
converter is adopted in practical engineering because it is
suitable for computer simulation of control system. It is mainly
used to replace the switching circuits in simulation environment to make the simulation run less time. Correspondingly
it is used conventionally for obtaining small signal models,
calculating consuming converter losses as well. It is created
using averaged switch modeling, each converter leg is treated
as switching cell, and is modeled as a pair of dependent
voltage and current sources, which represent the average
voltage and current generated by the switching cell over a
single switching cycle. In averaged circuit model, the purposes
are less complexity and faster time domain simulation, while
still maintaining sufficient converter dynamic accuracy. The
control circuit is modeled in the same way as if a full
switched model were used except that modeling of a gate
pulse generation is not needed. Furthermore, over-modulation,
saturation effects and other non-linearity’s are also modeled
correctly. On the other hand, the model is not suited for
analysis where consequences of switching frequency ripple
phenomena are focused. In fact, it is suitable for calculating
the converter losses in drives at any load and speed.
Fig. 1.
III. E LECTROTHERMAL M ODELING AND L IFETIME
E STIMATION OF THE VOLTAGE S OURCE C ONVERTER FOR
W IND T URBINE
System under study [7][8]
λds = Ls ids + Lm idr
(9)
λqs = Ls iqs + Lm iqr
(10)
λdr = Lm ids + Lr idr
(11)
λqr = Lm iqs + Lr iqr
(12)
Te =
3
pLm (iqs idr − ids iqr )
2
(13)
where:
Ls = Lls + Lm
(14)
Lr = Llr + Lm
(15)
The procedure for calculating the estimated lifetime for
semiconductors devices in wind energy applications can be
schematized in a logical steps sequence as shown in Fig. 2.
Started from calculating the losses, which is based on the look
up table method for calculating the conduction and switching
losses. An equivalent (RC) network model is built to perform
the thermal analysis, therefor determination of virtual junction
temperature, cycle counting and lifetime prediction according
Miner’s rule. The real time simulation environment dictates
the requirements for the models: easy implementation on the
software platform Simulink/Matlab and fast calculation time.
The parameters of the system under analysis are defined
as follows:
TABLE I.
DFIG E LECTRICAL PARAMETERS
P ower, Pn
7.5 KW
StatorV oltage, Vn
415 V
f requency, fn
50 Hz
Rs
1.06 Ω
Ls
0.2065 Ω
Rr
0.8 Ω
Lr
0.081 H
Lm
0.0644 H
Inertia J
7.5 kgm2
P ole pairs
3
Rating speed
970 rpm
Fig. 2.
Power Semiconductor Lifetime Estimation Model [10]
A. Power Losses of (IGBT) in the (RSC)
There are three kinds of losses in power devices: conduction losses (static losses), switching losses (transient losses)
and gate losses. Increasing the switching frequency will increase switching losses so that it become the dominant factor
in the total power losses, where the gate losses is left out as it
is in insignificant compared to the switching and conduction
losses.
A fast power losses simulation model for a three phase
inverter power module thermal simulation is proposed in this
paper, the fast accurate power losses simulation method is
implemented for power devices thermal simulation, due to
large simulation time steps applied, this allows power losses
and thermal performance of a device in an converter to
be predicted over long periods of real time, altogether this
simulation methodology brings together accurate models of
the electrical systems performance. The speed up is obtained
by simplifying the representation of three phase inverter at
the system modeling stage using large time step, average
model is used to calculate the power losses using predefined
look up table, this simulation methodology brings together
accurate models for of the electrical system performance, over
that suitable CPU time simulation for long real time thermal
simulation of inverter power device.
B. Thermal Modeling Technique
(RC) ladder networks are more popular to use for thermal
analysis, they are easy to integrate into existing circuit simulator making the latter capable of it simulating both electrical and
thermal characteristics of circuits, the (RC) thermal model is
flexible. Building equivalent thermal (RC) model is our choice
in real time simulator for its easy implementation and short calculation time [11]. To determine the values of the various “R”s
and “C”s would be to extract their values from the dynamical
thermal impedance curve available from experiment or from
simulation or from manufacture data sheet available as in table
II. Operating temperature play a major role in consequence
for performance and reliability of semiconductors devices, it
is not surprising that the safety margin or reliability of a
semiconductor devices decreases as the temperature increases.
The thermal RC circuits for (IGBT)is built, using the power
losses as current source value in the circuit, the junction and
case temperature can be determined for corresponding node
voltages.
C. Lifetime Prediction and Design of Reliability
In recent years there are various models and counting algorithms to estimate the lifetime of an IGBT power module, they
differ in the number of parameters used to specify a temperature cycle. Basically the lifetime estimation of power module
demands the linkage of an application typical load profile
with a module specific lifetime model by counting algorithm.
There are several cycle counting methods being developed, for
example, the level crossing counting method, the peak counting
method and the simple range/mean counting method. However,
these methods cannot capture all the characteristics needed for
accurate fatigue analysis. The Rainflow cycle counting method,
which was developed in 1968 by Endo and Matsuishi is one
of the most popular cycle counting technique used in fatigue
analysis. And used to extract closed loading cycles [12], the
origin of the name of Rainflow counting method is called
“Pagoda Roof Method”. The IGBT lifetime prediction models can split into analytical and physical models. Analytical
lifetime models estimate the life of the device in terms of
number of cycles to failure Nf considering various factors
such as temperature swing, medium temperature, frequency
and bond wire current. The main problem with analytical
lifetime models is that it is difficult to accurately extract
the number and amplitude of the temperature by Rainflow
counting algorithm[13]. operation over a spectrum of different
stress levels results in a damage fraction Di for each of the
different level stress levels Si in the spectrum. It is clear that,
failure occurs if the fraction exceeds unity as in equation 19.
When load profiles are unpredictable, consider the entire time
history of the load as input, where cycle counting algorithms
have been used to represent the spectrum of the load into a
set of simple uniform data histograms, its important is that it
allows the applications of Miner’s Rule [14], The B10 lifetime
models of the power devices [15] are to map the k th counted
thermal cycles to the number of cycles that the IGBT has 10%
failure rate, Nklif e . The consumed B10 lifetime by each cycle
CLk is calculated as reciprocal of Nklif e . Finally the Miners
rule is applied to calculate consumed life time by total number
of thermal cycles K during 30 seconds interval. The number N
of cycles until a certain percentage of the modules fail can be
calculated from the temperature excursion ∆T by the inverse
power law relationship [15]:
N = k1 .∆T −k2
where the two parameters k1 , k2 respectively are scale
parameter and the exponent parameter and both of them
are device dependent and have to be determined based on
measurements. According to Palmgren Miner linear damage
accumulation rule, the effects of different loads can be combined, and the life consumption LC or the damage fraction
at any stress level Si is linearly proportional to the ratio of
number of cycles of operation to the total number of cycles
that produces failure at that stress level, the accumulated
damage satisfied by [16] as in equation 18. The lifetime of
the IGBT is predicted to be 4818.5 hours if running at these
load conditions.
LC =
k
X
ni
Nf i
i
IGBT T HERMAL C HARACTERISTIC VALUES
i
1
2
3
4
Ri (K/kW )
35.1
8.25
3.85
3.79
τi (ms)
207.4
30.1
7.6
1.6
(17)
Then, a total damage can be defined as the sum of all the
fractional damages over a total of k blocks.
n at∆T2
n at∆T1
+
+ ...
N10%,∆T =∆T1
N10%,∆T =∆T1
n at∆Tk
+
< 100% (18)
N10%,∆T =∆Tk
or
D1 + D2 + ... + Di−1 + Di ≥ 1.0
TABLE II.
(16)
(19)
and the event of failure can be defined as:
D ≥ 1.0.
(20)
IV.
P RINCIPLES OF F ILTERING W IND T URBINE P OWER
C OMMAND F LUCTUATIONS
In modern power electronic in wind energy conversion
systems based on variable speed drives depends on power command fluctuation. One of the pivotal parameters for lifetime
estimation for IGBTs is the the thermal environment and the
number of thermo cycles the device undergo. Solid state devices have in general good reliability and long lifetime as there
are no moving parts involved that may wear out. However, the
device are fragile to excessive voltage and currents, and can be
damaged even by very short duration shocks above maximum
ratings. In well designed system, the solid state devices are
well protected from such events with little bluster to lifetime.
Studies have shown that the power cycling of the IGBT module
is one of the dominant failure mechanism of high power IGBT
multichip module [1]. A low pass filter LPF has been proved to
be an effective approach to suppress thermo cycles, the design
procedure and control consideration for this topology, to reduce
thermo cycles, the control strategy is mostly focused on wind
turbine side active power control but not the power grid side.
However, the wind turbine’s output power fluctuation due to
wind speed variations, therefore, a LPF used to smoothing the
fluctuations. The paper proposes a new wind power generation for which the smoothing performance is examined, the
simulation result shows that a new wind power generation has
an excellent smoothing performance for output power wind
turbines, proposed system has the advantage that the thermo
cycles of the inverter are less than those of the inverter of
a conventional system. The reference value of of the power
control for the conventional controller is the output of the
low pass filter, The output of the wind turbine is the input
to the low pass filter. The power fluctuation smoothing is
carried out by charging or discharging the difference between
the reference value and wind turbine power output. The time
constant of this filter is the range from several second to days.
thus the new wind power generation in which the reference
power fluctuation smoothing performance is given by a such a
a scheme, controlling the power output of the wind turbine
to track power set point commands. The traditional goal
of these wind power plants is to maximize profitability by
maximizing energy extraction, and therefore the power output
of the wind plants often varies with fluctuation winds, absorb
the fluctuation components of the wind power effectively,
and to reduce number of thermo cycles. A comparison of
energy production over lifetime and number of thermal cycling
between with and without LPFunder the same wind turbine
mission profile will be introduced in our final paper.
Fig. 4.
Power Command (W), dashed: with LPF, line:without LPF
Fig. 5.
DC Bus Voltage (V) without LPF
Fig. 6.
IGBT Temperature Variation (o C) without inserting LPF
Fig. 7.
Frequency distribution of temperature cycles defined by their
amplitude ∆T and temperature mean value Tm extracted from Rainflow
counting algorithm without inserting LPF
V.
Fig. 3.
Wind speed (m/s)
C ONCLUSION
The prediction of power cycling lifetime for a power
electronic converter in rotor side control in DFIG is examined.
A comprehensive thermal model for the power IGBT modules
used in three-phase converter is build in order to predict
the dynamic junction temperature rise under real operating
conditions. The power losses model, which is based on the
look-up table method for calculating the conduction and
switching losses are successfully simulated. An equivalent
RC network model is built to perform the thermal analysis.
The parameters of the thermal network are extracted from
the junction to case and case to ambient dynamic thermal
impedance curves. Lifetime is estimated. An averaged model
is suitable for electro thermal simulation, it can be used
to calculate the semiconductor losses at any output current
waveform, it considered faster and it can be parameterized
with conventional data sheet information. The analysis shows
that the lifetime look alike heavily influenced by thermal
cycling, and the behavior of the semiconductor devices and
their mission profile which directly affects the lifetime. Hence,
an adequate control for smoothing the power injected in the
DFIG and absorb some of the thermo cycles, where a low
pass filter LPF is inserted in the circuit, by using the proposed
method, the stress of the capacity and the stress on the
power converter are reduced, reducing power fluctuation cause
significant impact on number of thermo cycles of the IGBT
powered converter, with increasing of lifetime estimation. In
fact a typical junction temperature profile would contain both
low frequency and high frequency components due to variation
of wind speed, simulation results confirm that the number of
thermo mechanical power cycling stress are strongly affected
by LPF, in addition, results prove that lifetime consumption
are improved using LPF.
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
R EFERENCES
[1]
[2]
[3]
[4]
M. Ciappa, F. Carbognani, and W. Fichtner, “Lifetime prediction
and design of reliability tests for high-power devices in automotive
applications,” Device and Materials Reliability, IEEE Transactions on,
vol. 3, no. 4, pp. 191–196, Dec 2003.
L. Wei, R. Kerkman, R. Lukaszewski, H. Lu, and Z. Yuan, “Analysis of
igbt power cycling capabilities used in doubly fed induction generator
wind power system,” Industry Applications, IEEE Transactions on,
vol. 47, no. 4, pp. 1794–1801, July 2011.
K. Ma, A. Bahman, S. Beczkowski, and F. Blaabjerg, “Complete loss
and thermal model of power semiconductors including device rating
information,” Power Electronics, IEEE Transactions on, vol. 30, no. 5,
pp. 2556–2569, May 2015.
Z. Zhou, M. Khanniche, P. Igic, S. Kong, M. Towers, and P. Mawby, “A
fast power loss calculation method for long real time thermal simulation
of igbt modules for a three-phase inverter system,” in Power Electronics
and Applications, 2005 European Conference on, 2005, pp. 9 pp.–P.10.
[20]
[21]
[22]
[23]
[5]
T. P. P.H. Wirsching and K. Ortiz, “Random vibrations: Theory and
practice.” Courier Dover Publications, 2006.
[24]
[6]
L. Yang, G. Yang, Z. Xu, Z. Dong, K. Wong, and X. Ma, “Optimal
controller design of a doubly-fed induction generator wind turbine system for small signal stability enhancement,” Generation, Transmission
Distribution, IET, vol. 4, no. 5, pp. 579–597, May 2010.
[25]
[7]
[8]
[9]
C. Busca, R. Teodorescu, F. Blaabjerg, S. Munk-Nielsen, L. Helle,
T. Abeyasekera, and P. Rodriguez, “An overview of the reliability
prediction related aspects of high power {IGBTs} in wind
power applications,” Microelectronics Reliability, vol. 51, no.
911, pp. 1903 – 1907, 2011, proceedings of the 22th European
Symposium on the {RELIABILITY} {OF} {ELECTRON} DEVICES,
{FAILURE} {PHYSICS} {AND} {ANALYSIS}. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0026271411002526
R. Pena, J. Clare, and G. Asher, “Doubly fed induction generator using
back-to-back pwm converters and its application to variable-speed windenergy generation,” Electric Power Applications, IEE Proceedings -,
vol. 143, no. 3, pp. 231–241, May 1996.
. L. T. A. Novotny, D. W., “Vector control and dynamics of ac drives,”
in Oxford. Clarendon Press, 1996.
[26]
[27]
[28]
L. GopiReddy, L. Tolbert, B. Ozpineci, and J. Pinto, “Rainflow algorithm based lifetime estimation of power semiconductors in utility
applications,” Industry Applications, IEEE Transactions on, vol. PP,
no. 99, pp. 1–1, 2015.
Y. Yu, T.-Y. Lee, and V. Chiriac, “Compact thermal resistor-capacitornetwork approach to predicting transient junction temperatures of a
power amplifier module,” Components, Packaging and Manufacturing
Technology, IEEE Transactions on, vol. 2, no. 7, pp. 1172–1181, July
2012.
C.Lalanne, “Mechanical vibration and shock analysis, fatigue damage.”
John Wiley and Sons, 2014.
A. Niesłony, “Determination of fragments of multiaxial service loading
strongly influencing the fatigue of machine components,” Mechanical
Systems and Signal Processing, vol. 23, no. 8, pp. 2712–2721, 2009.
R. Amro, J. Lutz, and A. Lindemann, “Power cycling with high temperature swing of discrete components based on different technologies,” in
Power Electronics Specialists Conference, 2004. PESC 04. 2004 IEEE
35th Annual, vol. 4, 2004, pp. 2593–2598 Vol.4.
N. Kaminski, “Load Cycle Capability of HiPaks, Application Notes
5SYA 2043-01,” ABB Switzerland Ltd, Semiconductors, Tech. Rep.,
Sep 2004.
M. Musallam and C. Johnson, “An efficient implementation of the rainflow counting algorithm for life consumption estimation,” Reliability,
IEEE Transactions on, vol. 61, no. 4, pp. 978–986, Dec 2012.
J. Kolar and F. Zach, “Losses in pwm inverters using igbts,” Electric
Power Applications, IEE Proceedings -, vol. 142, no. 4, pp. 285–288,
Jul 1995.
S. Clemente, “Transient thermal response of power semiconductors to
short power pulses,” Power Electronics, IEEE Transactions on, vol. 8,
no. 4, pp. 337–341, Oct 1993.
S. Azuma, M. Kimata, M. Seto, X. Jiang, H. Lu, D. Xu, and L. Huang,
“Research on the power loss and junction temperature of power semiconductor devices for inverter,” Vehicle Electronics Conference, 1999.
(IVEC ’99) Proceedings of the IEEE International, pp. 183–187 vol.1,
1999.
T. Reimann, R. Krummer, U. Franke, S. Petzoldt, and L. Lorenz, “Real
time calculation of the chip temperature of power modules in pwm
inverters using a 16 bit microcontroller,” Power Semiconductor Devices
and ICs, 2000. Proceedings. The 12th International Symposium on, pp.
127–130, 2000.
A. Ammous, K. Ammous, M. Ayedi, Y. Ounajjar, and F. Sellami, “An
advanced pwm-switch model including semiconductor device nonlinearities,” Power Electronics, IEEE Transactions on, vol. 18, no. 5, pp.
1230–1237, Sept 2003.
D. Murdock, J. Torres, J. Connors, and R. Lorenz, “Active thermal
control of power electronic modules,” Industry Applications, IEEE
Transactions on, vol. 42, no. 2, pp. 552–558, March 2006.
A. Isidoril, F. Rossi, F. Blaabjerg, and K. Ma, “Thermal loading
and reliability of 10-mw multilevel wind power converter at different
wind roughness classes,” Industry Applications, IEEE Transactions on,
vol. 50, no. 1, pp. 484–494, Jan 2014.
A. Hangleiter, A. Grabmaier, and G. Fuchs, “Anomalous damping in
mqw lasers due to slow inter-well transport,” Electron Devices, IEEE
Transactions on, vol. 40, no. 11, pp. 2106–, Nov 1993.
K. Ma, M. Liserre, F. Blaabjerg, and T. Kerekes, “Thermal loading
and lifetime estimation for power device considering mission profiles
in wind power converter,” Power Electronics, IEEE Transactions on,
vol. 30, no. 2, pp. 590–602, Feb 2015.
L. Wei, J. McGuire, and R. Lukaszewski, “Analysis of pwm frequency
control to improve the lifetime of pwm inverter,” Industry Applications,
IEEE Transactions on, vol. 47, no. 2, pp. 922–929, March 2011.
L. GopiReddy, L. Tolbert, and B. Ozpineci, “Lifetime prediction of igbt
in a statcom using modified-graphical rainflow counting algorithm,” in
IECON 2012 - 38th Annual Conference on IEEE Industrial Electronics
Society, Oct 2012, pp. 3425–3430.
M.
Ciappa,
“Selected
failure
mechanisms
of
modern
power
modules,”
Microelectronics
Reliability,
vol.
42,
no. 45, pp. 653 – 667, 2002. [Online]. Available:
http://www.sciencedirect.com/science/article/pii/S0026271402000422