electronics

Article
Thermal Safety Assessment Method for Power Devices in
Natural Air-Cooled Converters
Weichao He 1 , Yiming Zhu 1, * , Zijian Liu 1 , Jiaxue Lin 1 , Fengshuo Bao 1 , Wei Zu 2 and Zhanfeng Ying 1, *

1 School of Energy and Power Engineering, Nanjing University of Science and Technology,
Nanjing 210094, China; [email protected] (W.H.); [email protected] (Z.L.);
[email protected] (J.L.); [email protected] (F.B.)
2 National Key Laboratory of Transient Physics, Nanjing University of Science and Technology,
Nanjing 210094, China; [email protected]
* Correspondence: [email protected] (Y.Z.); [email protected] (Z.Y.)

Abstract: The junction temperature of a power device in a natural air-cooled power converter
fluctuates randomly due to the variation in airflow rate in ambient environments. Most of the
existing thermal analysis methods do not pay attention to the uncertain influence of airflow on the
heat-dissipation capacity of such converters, making it difficult to accurately evaluate the thermal
safety of these devices. To address this issue, a thermal safety assessment method for power devices
in natural air-cooled converters is proposed in this paper. In the proposed method, convective heat
resistance samples of converter housing are extracted with an equivalent thermal network model
and the historical operation temperature of the converter. Wavelet packet transform is used to
analyze the time–frequency domain characteristics of the convective heat resistance, and Monte
Carlo simulation is employed to simulate the random influence of the airflow rate on the device
junction temperature. The thermal safety of power devices is assessed in the form of over-temperature
probability, which is expressed by a two-variable growth function. An experimental platform is
designed to validate the effectiveness of the proposed method. The results show that the proposed
method can accurately estimate the over-temperature risk of a power device in a natural air-cooled
converter under different ambient temperature and current levels, thus effectively improving the
thermal reliability of converters.
Citation: He, W.; Zhu, Y.; Liu, Z.; Lin,
J.; Bao, F.; Zu, W.; Ying, Z. Thermal
Keywords: power converter; thermal safety; natural air cooled; wavelet packet transform; Monte
Safety Assessment Method for Power Carlo simulation; growth function
Devices in Natural Air-Cooled
Converters. Electronics 2024, 13, 1665.
https://doi.org/10.3390/
electronics13091665 1. Introduction
Academic Editor: Niamat Hussain Power devices are widely used in power electronic converters, such as power supplies,
motor drivers and grid-connected inverters. To avoid overheating fault in the converter, the
Received: 18 March 2024 junction temperature of the power device must not exceed its maximum allowable value.
Revised: 21 April 2024
Therefore, how to accurately assess the thermal safety of power devices under different
Accepted: 23 April 2024
working conditions is a key issue to ensure the reliability of converters.
Published: 25 April 2024
The thermal safety assessment of devices relies on effective thermal analysis methods.
At present, the commonly used device thermal analysis methods mainly include the finite
element method and the equivalent thermal network model method [1]. The finite element
Copyright: © 2024 by the authors.
method simulates the 2D or 3D temperature field distribution of a power device and its
Licensee MDPI, Basel, Switzerland. converter housing through the grid division of the calculation area [2,3]. Due to the huge
This article is an open access article amount of calculation required, the finite element method is more suitable for design-
distributed under the terms and ers of semiconductor devices who require detailed information about the time–spatial
conditions of the Creative Commons temperature distribution of a device structure. The equivalent thermal network model
Attribution (CC BY) license (https:// estimates variations in device temperature and converter housing temperature based on
creativecommons.org/licenses/by/ the thermal–electrical analogy [4]. The equivalent thermal network model is expressed as
4.0/). an RC electrical network based on the thermo-electric analogy in which the voltage, current,

Electronics 2024, 13, 1665. https://doi.org/10.3390/electronics13091665 https://www.mdpi.com/journal/electronics

Electronics 2024, 13, 1665 2 of 19

and RC parameter correspond to temperature, thermal power, and thermal impedance,

respectively. Since the thermal network model only provides the temperature information
of a limited number of nodes (mainly junction, case and heatsink temperature) between the
device and the environment, it has the advantages of small calculation requirements and
convenient use. For this reason, the thermal network model is usually used for the online
thermal analysis of power devices.
The above two thermal analysis methods have been widely used for various power
devices and converters. The finite element method was employed in the literature [5] to
establish a 3D thermal model of IGBT and to analyze the sensitivity of the model param-
eters. The authors of [6,7] used the finite element method to simulate the electrothermal
performance of IGBT and successfully realize the retrieval of voltage and current. The
thermal network model was used to estimate the junction temperature of power devices
and realize junction temperature control to improve the overload capacity of converters [8].
The authors of [9] proposed a thermal network model of power, MOSFET, which consid-
ers the thermal aging effect of packaging materials and accurately estimates the junction
temperature change during the continuous degradation of a device. Based on a thermal
network model of power devices, the life of a wind turbine converter under a long-term
task profile was evaluated in [10]. Due to the aging of device materials and nonlinear heat
dissipation, the heat impedance in the thermal network model may change nonlinearly
with use time and working conditions. In order to reflect the junction temperature change
in devices more accurately, the nonlinear variation characteristics of thermal impedance
in thermal network models are studied in the literature [11,12]. These nonlinear network
models considering temperature-dependent physical parameters have been proven to
have higher thermal analysis accuracy [13,14]. In the literature [15,16], three-dimensional
thermal network models are constructed to estimate the internal temperature distribu-
tion of modules composed of multiple chips. Compared with one-dimensional models,
three-dimensional thermal models have significant advantages in describing the thermal
coupling effects between device chips [17–19].
Although the above studies have greatly promoted the development of thermal analy-
sis technology for power devices and converters, they assume that the research object works
under stable and deterministic conditions. Consequently, the existing thermal analysis
methods are not suitable for assessing the thermal safety of devices in natural convection
air-cooled converters. A natural air-cooled converter is a kind of converter that uses the
natural flow of air in an ambient environment to dissipate heat. Typical air-cooled convert-
ers include portable DC power supplies, movable inverters, photovoltaic controllers, etc.
These converters are commonly low-power, but play a crucial role in the field of outdoor
power conversion and transmission. To meet waterproof and heat-dissipation performance
requirements, natural air-cooled converters usually have a fully enclosed metal housing.
Due to the uncertainty of ambient airflow rate, the heat-dissipation capacity of these con-
verters is unstable, which leads to random fluctuations in device junction temperature. The
randomness of the junction temperature greatly increases the difficulty of thermal safety
assessment, thus reducing the effectiveness of thermal management strategies. Therefore, it
is still a great challenge to accurately evaluate the thermal safety of power devices in natural
air-cooled converters because existing studies commonly ignore the uncertain influence of
airflow rate on converters.
To address this problem, a thermal safety assessment method is proposed in this
paper. In the proposed method, the random effect of airflow rate on natural air-cooled
converters is quantitatively analyzed by a wavelet packet transform on the convective
heat resistance of the converter housing. Considering the advantage of online usability,
the equivalent thermal network model of the converter is used to extract the convective
heat resistance samples of the housing. The Monte Carlo method is employed to simulate
random fluctuation in the device’s junction temperature. A two-variable growth function is
used to describe the over-temperature probability of the device under different current and
ambient temperatures. Compared with existing thermal analysis studies, this study can
Electronics 2024, 13, 1665 3 of 19

effectively reflect the thermal safety of a device in the form of over-temperature probability,
thus improving the reliability of natural air-cooled converters.

2. Influence of Airflow Rate Randomness on the Thermal Safety of Natural

Air-Cooled Converters
2.1. Equivalent Thermal Network Model of Natural Air-Cooled Converters
As described in the introduction, in order to analyze the thermal safety of power
devices in practical engineering, it is necessary to estimate the device junction temperature
with the equivalent thermal network model. The structure of the thermal network model is
related to the physical structure of converters. Natural air-cooled converters commonly
have a fully enclosed structure and power devices are tightly attached to the inner wall
of the converter housing for heat dissipation. Due to the conductive heat resistance of the
converter housing, the power device may cause a certain degree of temperature distribution
on the housing. According to the physical definition [20], conductive heat resistance θ d can
be expressed as
l
θd = (1)
λS
where λ is the thermal conductivity, which is determined by the material properties of the
housing. l is the path length of the heat conduction on the housing, which is related to the
distribution distance of power devices and the size of the housing. S is the heat conduction
area on the housing, which is related to the size and thickness of the housing.
In general, the smaller the converter housing and the higher the thermal conductivity
of the housing material, the smaller the value of the conductive heat resistance θ d . If θ d
is small enough, the temperature distribution on the converter housing can be ignored
and the housing can be regarded as a whole for thermal analysis. Due to the lack of an
auxiliary cooling system, the housing of natural air-cooled converters must be made of
materials with good thermal conductivity, such as aluminum or copper. In addition, under
the requirements of cost and portability, the housing size of natural air-cooled converters
is commonly strictly limited. Because of the above two reasons, the housing temperature
distribution of many natural air-cooled converters in practical engineering is small.
On the basis of ignoring housing temperature distribution, the equivalent thermal
network model of a natural air-cooled converter composed of power MOSFETs can be
constructed as shown in Figure 1. The equivalent thermal network model consists of a
thermal network from the device junction to the converter housing and a thermal circuit
from the converter housing to the ambient environment. The subscript n represents the
number of power devices inside the converter. The parameters Ploss,n , Tj,n and Tc,n are
the loss, junction temperature and case temperature of the device, respectively. Ci ,n , θ i ,n
and Ti ,n are the heat capacity, heat resistance and temperature of the i-node (i from 1 to m)
inside the device, respectively. Cce,n and θ ce,n are the heat capacity and heat resistance of
the heat-conducting material between the device case and the converter housing, respec-
tively. Te , and Ta are the converter housing temperature and the ambient temperature,
respectively. Ce and θ e are the heat capacity and convection heat resistance of the converter
housing, respectively.
According to the electrical network analysis method, the thermal networks from the
device junction to the converter housing can be expressed as
dTj,n Tj,n − T1,n

C1,n dt = Ploss,n − θ1,k




Tj,n − T1,n T1,n − T2,n

 dT1,n
C2,n dt = θ1,n − θ2,n




...... (2)

− Tm−1,n −T

dT T T
Cm,n mdt−1,n = m−2,n − m−θ1,nm,n c,n





 θm−1,n

−T

dT T T − Te
Cce,n dtc,n = m−θ1,nm,n c,n − c,n


θce,n
 Electronics 2024, 13, 1665 4 of 19

The thermal circuit from the converter housing to the ambient environment can be
expressed as
Tc,n − Te Te − Ta
 
dTe
Ce =∑ − (3)
dt k
θ ce,n θe
When the converter housing temperature Te is measured, the case temperate Tc,k and
the junction temperature Tj,k can be directly estimated by (2). If the measured temperature is
Electronics 2024, 13, 1665 4 of 19
the ambient temperature Ta , the junction temperature needs to be estimated by combining
(2) and (3).

Ploss,n Ploss,2 Ploss,1

device junction
Tj,n Tj,2 Tj,1

C1,n …… C1,2 C1,1

θ1,n θ1,2 θ1,1
T1,n T1,2 T1,1
C2,n θ 2,n C2,2 C2,1 θ 2,1
θ 2,2
…

…
…

…
……
Ti −1,n Ti −1,2 Ti −1,1
Ci ,n Ci ,2 Ci ,1
θi ,n θi ,2 θi ,1
…

…
…

Tm −1,n Tm −1,2 Tm −1,1

Cm , n θ m,n Cm,2 Cm,1
θ m,2 θ m ,1
……
Tc,n Tc,2 Tc ,1 converter housing
Cce,n Cce,2 Cce ,1
θ ce,n θ ce,2 θ ce,1
Te

Ce

θe
ambient environment
Ta

Figure1.1.Thermal
Figure Thermalnetwork
networkmodel
modelof of
a natural air-cooled
a natural converter.
air-cooled converter.

Obviously,
Accordingthe device
to the junction
electrical temperature
network analysis is estimated
method, the withthermal
the linear thermalfrom
networks net- the
work in this paper. However, as mentioned in the introduction, the heat resistances in the
device junction to the converter housing can be expressed as
thermal network have certain nonlinear characteristics. The parameter extraction workload

of the nonlinear thermal network disTj,larger
n
than T j, n − T1, n
that of the linear thermal network, so the
C1,widely
linear thermal network is still n =used n −estimate the device junction temperature
Ploss,to
dt
 the continuous θ1, k
in practical engineering. With improvement and development of parame-

ter identification methods, nonlinear Tj, n − networks
dT1, nthermal T1, n T1, n −can T2, n be used to construct thermal
C2, n = −
models of natural air-cooled convertersdt in the θ future. θ 1, n 2, n

2.2. Experimental Set-Up  ...... (2)
 dT
To analyze the influence of airflow T randomness
rate −T − T thermal safety of the
T on the
C , n m −1, n = m − 2, n m −1, n − m −1, n c, n
natural air-cooled converter,anmexperimental
dt θ m −1, n
platform is θ m, n as shown in Figure 2.
designed,
In this platform, an H-bridge converter with output square-wave voltage is used as the
experimental object. The loadCof the n Tm −1, n −isTac, n0.5TΩ
dTc,converter − Te
c, n resistor and a 0.1 mH inductor.
 ce, n = −
An 18 V DC power supply is used dto
 t powerθthe θce, n The circuit of the converter
m , n converter.
is mounted inside a rubber base and a metal housing, as shown in Figure 3. The five
Theofthermal
surfaces circuit
the housing are from the converter
numbered housing to the
S1 to S5, respectively. The ambient
size of theenvironment can be
housing is only
100 mm × 32
expressed as mm × 32 mm. The material of the housing is 1070 aluminum alloy with a
thermal conductivity of up to 234 W/(m·k).
dTe  T −T  Te − Ta
Ce =  c, n e  − (3)
dt k  θ θe
 ce, n 
When the converter housing temperature Te is measured, the case temperate Tc,k and
the junction temperature Tj,k can be directly estimated by (2). If the measured temperature
is the ambient temperature Ta, the junction temperature needs to be estimated by combin-
ing (2) and (3).
Obviously, the device junction temperature is estimated with the linear thermal net-
work in this paper. However, as mentioned in the introduction, the heat resistances in the
 perimental object. The load of the converter is a 0.5 Ω resistor and a 0.1 mH inductor. An
perimental
of18the object.
areThe load of the converter is a 0.5 Ω The
resistor
sizeand a 0.1 mH inductor. An mm
V housing
DC power numbered
supply S1 to
is used S5,power
to respectively.
the converter. of the
The housing
circuit is only
of the 100
converter is
18 V DC
×mounted power
32 mm ×inside supply
32 mm. The is used to
material power
of the the converter.
housing is 1070The circuit
aluminum of the
alloyconverter
with a is
thermal
a rubber base and a metal housing, as shown in Figure 3. The five surfaces
mounted inside a rubber base and a metal housing, as shown in Figure 3. The five surfaces
conductivity
of the housingof are
up to 234 W/(m·k).
numbered S1 to S5, respectively. The size of the housing is only 100 mm
of the housing are numbered S1 to S5, respectively. The size of the housing is only 100 mm
Electronics 2024, 13, 1665
× ×3232mmmm × 32
× 32 mm.mm.The The material
material of the
of the housing
housing is 1070
is 1070 aluminum
aluminum alloyalloy
with with
5 of a
19thermal
a thermal
conductivity of up to 234 W/(m·k).
conductivity of up to 234 W/(m·k). Load PC
Internal circuit of
power converter

Load Load PC PC
Internal
Internal circuit
circuit of of
power converter DC power supply
power converter

Controller DCsupply
DC power power supply

Convective environment Converter

Electric fan

Controller
Controller

Convective environment Converter

Converter
Electric fan fan
Electric Convective environment
Figure 2. Experimental platform.

Figure 2. 2.
Figure
Figure Experimental
2. platform.
Experimental
Experimental Top Surface (S2)
platform.
platform. Right surface (S4)
Back surface (S1)
Top Surface (S2)
Top Surface (S2) Right surface (S4)
Back surface (S1) Right surface (S4)
Back surface (S1) Metal
housing
Left surface (S3) Metal
Front surface (S5) housingMetal
housing
Left surface (S3)
Left surface (S3) Front surface (S5)
Front surface (S5)

Figure 3. Schematic diagram of the converter structure in the experimental platform.

Figure 3. 3.
Figure Schematic
Schematicdiagram
diagramof
ofthe
theconverter structurein
converter structure inthe
theexperimental
experimentalplatform.
platform.
The3. converter
Figure consistsofofthe
Schematic diagram four powerstructure
converter MOSFETs numbered
in the fromplatform.
experimental Q1 to Q4, as shown
in Figure
The 4. The type
Theconverter
converter of power
consists
consists of four
of of power
four the MOSFETs
power MOSFETs
MOSFETs isnumbered
IRFB4410PbF,
numberedfrom and
fromQ 1Qtothey
Q , asare attached to
shown
1 to 4Q4, as shown

inin
the Figure
inner
Figure 4.
The4.wallThe
of
converter
The type
typetheof of power
housing
consists
power ofof the
through
four
the MOSFETs
power
MOSFETs isisIRFB4410PbF,
the thermal
MOSFETs IRFB4410PbF,
conductive
numberedandthey
theyare
adhesive
and from are attached
TR6400.
Q to Q4,The
1 attached ther-
astoshown
to
mal the inner wall
conductivity of the
and housing
thepower through
thickness the thermal
ofMOSFETs
the thermal conductive
conductive adhesive
adhesive TR6400.
are The
0.8attached
W/(m·k)to
ininner
the Figure 4. The
wall of thetype of
housing through of the
the thermal is IRFB4410PbF,
conductive adhesiveand they
TR6400. are
The ther-
thermal
and 0.2 conductivity
mm. Table and
1 thethe
lists thickness
thermal of the thermal conductive
parameters of the adhesive
power are 0.8 W/(m
MOSFETs, the·k)
thermal
mal conductivity and the thickness of the thermal conductive
the inner wall of the housing through the thermal conductive adhesive TR6400. The adhesive are 0.8 W/(m·k) ther-
and 0.2 mm. Table 1 lists the thermal parameters of the power MOSFETs, the thermal
conductive
and
mal mm.adhesive
0.2conductivity
Table 1and and
liststhe the
the housing,parameters
thermal
thickness ofwhich
the are of
thermalprovided
the by the
power
conductive manufacturers.
MOSFETs,
adhesive the
are thermal
0.8 W/(m·k)
conductive adhesive and the housing, which are provided by the manufacturers.
conductive
and 0.2 mm. adhesive
Tableand the housing,
1 lists the thermal which are provided
parameters by the
of the manufacturers.
power MOSFETs, the thermal
conductive adhesive and the housing, which are provided by the manufacturers.

Figure4.
Figure Main power
4. Main power topology
topologyofof
thethe
converter.
converter.
Figure 4. Main power topology of the converter.
Table 1. Thermal network model parameters of the power device.
Figure 4. Main power topology of the converter.
Parametric Numerical Value Parametric Numerical Value
θ 1,n /(◦ C/W) 0.2736 C1,n /(J/◦ C) 0.0014
θ 2,n /(◦ C/W) 0.3376 C2,n /(J/◦ C) 0.0123
θe /(◦ C/W) 1.0521 Ce /(J/◦ C) 12.2064
 θ1,n/(°C/W) 0.2736 C1,n/(J/°C) 0.0014
θ2,n/(°C/W) 0.3376 C2,n/(J/°C) 0.0123
θe/(°C/W) 1.0521 Ce/(J/°C) 12.2064
Electronics 2024, 13, 1665 6 of 19
A controller is used to control the load current of the converter and make the con-
verter commutate at a frequency of 200 Hz. The controller is designed based on an RK3568
microprocessorAwhose main
controller frequency
is used is up
to control thetoload
1.8 current
GHz/4 of core
theand the memory
converter and make is the
2 GB.
converter
To simulatecommutate at a variation
the uncertain frequencyinofairflow
200 Hz.rate The incontroller is designed
the experimental based on an
environment, theRK3568
microprocessor
controller controls whose
the motor speedmain frequency
of an is upbytothe
electric fan 1.8duty
GHz/4 core
cycle ofand the memory
the PWM signal.is 2 GB.
To simulate the uncertain variation in airflow rate
The control program in the controller is written based on C language. in the experimental environment, the
controller controls the motor speed of an electric fan by the duty
During the experiment, the temperature, including the device case temperature, the cycle of the PWM signal.
The control
converter housing program in
temperature thethe
and controller
ambientistemperature,
written basedare on measured
C language.by MF52 ther-
During the experiment, the temperature, including the device case temperature, the
mistors with an error of less than 1% and a thermal response time of less than 2 s. The
converter housing temperature and the ambient temperature, are measured by MF52
current is measured by an MT9221 Hall effect sensor with an error of less than 2%. All the
thermistors with an error of less than 1% and a thermal response time of less than 2 s. The
measurements areiscollected
current measured bybythe
ancontroller
MT9221 Hall andeffect
sent sensor
to a PCwith
for storage
an error and analysis.
of less than 2%. All the
measurements are collected by the controller and sent to a PC for storage and analysis.
2.3. Analysis on the Influence of Thermal Safety
2.3. Analysis
To accurately on thethe
simulate Influence of air-cooling
natural Thermal Safety
conditions of the converter, the airflow
rate with a totalToduration
accurately ofsimulate
180 minthewas natural air-cooling
measured in an conditions of the converter,
outdoor natural environment the airflow
rate with
with a sampling a total
period of duration of 180 min
1 s. According was
to the measured
measured in ansimilar
data, outdoor natural
airflow environment
changes
withina sampling
are generated period ofenvironment
the experimental 1 s. Accordingby to the
the measured data, similar
fan. A comparison of airflow changes are
the airflow
rate in the natural environment and the experimental environment is shown in Figure 5. rate in
generated in the experimental environment by the fan. A comparison of the airflow
the natural
It can be found environment
that the andgenerated
airflow rate the experimental environment
by the fan is in good is agreement
shown in Figure
with 5.the
It can be
found that the airflow rate generated by the fan is in good agreement with the airflow rate
airflow rate measured outdoors. The maximum full-scale deviation between the two is
measured outdoors. The maximum full-scale deviation between the two is only 9.1%F.S,
only 9.1%F.S, which indicates that the experimental platform can truly simulate the ran-
which indicates that the experimental platform can truly simulate the random variation in
dom variation in airflow rate in the natural air-cooling environment.
airflow rate in the natural air-cooling environment.

2.0

1.5
Airflow rate(m/s)

1.0

0.5

0.0
0 20 40 60 80 100 120 140 160 180
t(min)

(a)
2.0

1.5
Airflow rate(m/s)

1.0

0.5

0.0
0 20 40 60 80 100 120 140 160 180
t(min)

(b)
Figure 5. Cont.
 Electronics 2024, 13, 1665 7 of 19
lectronics 2024, 13, 1665
Electronics 2024, 13, 1665 7 of 19
7 of 19

10
10
8
8

Error (%)
6

Error (%)
6
4
4
2
2
0
0 0 20 40 60 80 100 120 140 160 180
t(min)
0 20 40 60 80 100 120 140 160 180
t(min)
(c)
(c)
Figure 5. Comparison of airflow rate in the natural environment and the experimental environment.
Figure
Figure 5.(a) 5. Comparison
Comparison
Airflow rate in theofoutdoor
of airflow airflow
rate ratenatural
in the in the natural
environment; (b) environment
environment
airflowand and
ratethe
in the experimental
experimental
experimental environment.
environment.
environment; (c) full-
(a) Airflow
(a) Airflow
scalerate rateoutdoor
in the
deviation in the
in outdoor
rate environment;
environment;
airflow between (b)the (b) airflow
airflow
two rate inrate in experimental
experimental
environments. environment;
environment; (c) full-
(c) full-
scale deviation in airflow
scale deviation rate between
in airflow the twothe
rate between environments.
two environments.
Figure 6 shows the junction temperature variation in the power device Q1 when the
Figure 6Figure
shows
converter
6 the
shows
operates
the junction temperature
junction
at a 7.5temperature
A load current
variation
variation theinpower
and theinabove
the power
random
device
device Q1 when
Q1 when
airflow
the
theconven-
rate. To
converter
converter operates operates
at athe at a 7.5
7.5junction A load
A load current current
and the and the above
above random random airflow
airflowunder rate.
rate. To To conve-
conven-
iently observe temperature fluctuation in the device natural convection
niently observe the junction temperature fluctuation in the device under natural convection
iently observe the
conditions, junction
the temperature
junction fluctuation
temperature in
presented the device
here is under
the natural
peak value convection
in each converter
conditions, the junction temperature presented here is the peak value in each converter
conditions, the junction
commutation. temperature presented here is the peak valuecaused
in eachbyconverter
commutation. That is, the fluctuation in junction temperature caused by commutation is is
That is, the fluctuation in junction temperature commutation
commutation.
not That is,in
presented the
thefluctuation
results. in junction temperature caused by commutation is
not presented in the results.
not presented in the results.
120

120
100
Junction temperature(℃)

100
Junction temperature(℃)

80

80
60

60
40

40
20
0 20 40 60 80 100 120 140 160 180
20
t(min)
0 20 40 60 80 100 120 140 160 180
t(min)
Figure 6. Junction
Figure 6. Junctiontemperature
temperaturevariation
variationinin the
the power
power device
device Q1Q1 when
when the the converter
converter operates
operates at a at a
Figure 6.7.5
Junction
A load temperature
current with variation
the randomin the power
airflow
7.5 A load current with the random airflow rate. device
rate. Q1 when the converter operates at a
7.5 A load current with the random airflow rate.
Due
Due to to the
theuncertainty
uncertaintyofofthe theairflow
airflow ininthe environment,
the environment, thethe
junction
junctiontemperature
temperature of of
Duetheto the
device uncertainty
Q also hasof the
strong airflow in
randomness, the environment,
even if the the
converter
the device Q11 also has strong randomness, even if the converter current is constant. In this junction
current temperature
is constant. of
In this
the device Q1 also
paper,
paper, the has
themaximum strongallowable
maximum randomness,
allowablejunction eventemperature
junction iftemperature
the converter Tjmax current
Tjmax of ofthetheis device
constant.
device In thisto beto be
is assumed
is assumed
◦
100maximum
C. It
paper, the
100 °C. It can
can bebe found
foundthat
allowable that the
therandom
junction junction
temperature
random junction temperature
Tjmax of the device
temperature causes
causestheassumed
is over-temperature
the to be
over-temperature
of
100 °C. It the
can device
be found in some
that periods.
the random The duration
junction of these
temperature
of the device in some periods. The duration of these periods is also uncertain, periods
causes is also
the uncertain, which
over-temperature greatlywhich
increases
of the device the difficulty of thermal safety assessment. The junction temperature of other
greatlyinincreases
some periods. The duration
the difficulty of thermal of safety
these periods
assessment. is also
Theuncertain, which
junction temperature of
power devices
greatly increases the has similar
difficulty of changes,
thermal which
safety are not described
assessment. The here.
junction temperature of
other power devices has similar changes, which are not described here.
other power In In
addition,
devices
addition,has thethe center
similar
center
temperature
changes,
temperaturewhichof of
are the
thenot five
five
metal surfaces
described
metal surfaceshere. is is also measured to
also measured to ver-
verify the good thermal conductivity of the converter housing, as shown in Figure 7a.
In addition,
ify thethermal
center temperature ofofthe five metal surfaces is as
also measured to ver-
Figure 7b shows the full-scale deviation between the temperature on theinsurface
the good conductivity the converter housing, shown Figure 7a.and
S1 Figure
ify the good
7b thermal conductivity of the converter housing, as shown in Figure 7a. Figure
theshows
temperaturethe full-scale deviation
on the other between
surfaces. the temperature
It is easy to find that the ontemperature
the surface S1 onand the tem-
the back
7b shows the
perature full-scale
on the deviation
other between
surfaces. It isthe
easytemperature
to find thatonthe
surface (S1) of the housing with the power device is the highest, while the temperature
the surface
temperature S1 and
on the
the tem-
back surface
perature(S1)
on the other surfaces. It is easy to find that the temperature
of the front surface (S5) is the lowest. The maximum full-scale deviation in temperaturefront
of the housing with the power device is the highest, while on
the the back
temperature surfaceof the
(S1) of the housing
surface
between (S5) with
is the
surface S1the power
lowest.
and The
surface device is the
maximum
S5 does nothighest,
full-scale while
exceed 8.5%F.S, the
deviation temperature
in temperature
indicating of the
that the front
between
temperature sur-
surface face
(S5) is
difference the lowest.
S1 andbetweensurface S5The maximum
does
these not exceed
surfaces full-scale
8.5%F.S,
is small. deviation in
indicating
Therefore, temperature conductivitysur-
between
that the temperature
the thermal difference
of the
face S1 and surface
converter
between theseS5 surfaces
housingdoesisnot goodexceed
is and 8.5%F.S,
small. its surface
Therefore, indicating
temperature
the thermal thatconductivity
the temperature
distribution canthe
of difference
beconverter
ignored in hous-
betweening these
this surfaces
study. is small. Therefore, the thermal conductivity
is good and its surface temperature distribution can be ignored in this study. of the converter hous-
ing is good and its surface temperature distribution can be ignored in this study.
Electronics 2024, 13, 1665 8 of 19
Electronics 2024, 13, 1665 8 of 19

120 12
S1 full scale deviation of temperature between S1 and S2
S2 full scale deviation of temperature between S1 and S3
S3 10 full scale deviation of temperature between S1 and S4
100 S4 full scale deviation of temperature between S1 and S5

Junction temperature(℃)
S5

Deviation（%）
8
80

60
4

40
2

20 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
t(min) t(min)

(a) (b)

Figure
Figure 7. Temperatures
7. Temperatures measured
measured onon
thethe different
different surfaces
surfaces of of
thethe housing
housing and
and thethe full-scale
full-scale devia-
devia-
tion between
tion between thethe
temperature
temperature onon
surface S1S1
surface and thethe
and temperature
temperature onon
other surfaces.
other (a)(a)
surfaces. Measured
Measured
temperature;
temperature;(b)(b)
full-scale deviation.
full-scale deviation.

3. 3. Time–Frequency
Time–Frequency Domain
Domain Analysis
Analysis forfor Convective
Convective Heat Heat Resistance
Resistance ofof Natural
Natural
Air-Cooled Converters
Air-Cooled Converters
3.1. Extracting Convective Heat Resistance of Converter Housing
3.1. Extracting Convective Heat Resistance of Converter Housing
Convective heat resistance is an important physical quantity for thermal analysis,
Convective heat resistance is an important physical quantity for thermal analysis,
since it can characterize the convective heat transfer intensity between a solid surface and
since it can characterize
an ambient environment the [21,22].
convective heat transfer
In order to quantify intensity betweenof
the influence a solid surface
airflow rate onandthe
anheat-dissipation
ambient environment capacity [21,22].
of theInconverter
order to quantify housing the andinfluence
the junction of airflow rate onof
temperature thethe
heat-dissipation capacity of the converter housing and
device, it is necessary to extract the convective heat resistance of the housing and analyze the junction temperature of the
device, it is necessary
its characteristics. to extracttothe
According heat convective
transfer theory, heat resistance
convective of heat
the housing
resistance and
hasanalyze
complex
itstime-varying
characteristics. characteristics because it is related to many nonlinear factors suchhas
According to heat transfer theory, convective heat resistance com-
as ambient
plex
wind speed, density, dynamic viscosity, and atmospheric pressure [23]. As a result, as
time-varying characteristics because it is related to many nonlinear factors such it is
ambient
difficult wind speed, density,
to accurately calculate dynamicconvective viscosity, heat and resistanceatmospheric pressuremethods.
by theoretical [23]. As aInre-
this
sult, it is difficult
paper, to accurately
the samples of convective calculate convectiveare
heat resistance heat resistance
extracted by theoretical
using the equivalent methods.
thermal
Innetwork
this paper, the samples of convective heat resistance
model and the historical operation temperature of the converter. are extracted using the equivalent
thermalBy network model
discretizing at and
timethe k, (2)historical
and (3) operation
are rewritten temperature
as of the converter.
By discretizingat time k, (2) and (3) are rewritten as
k −Tk
 
k + 1 ∆t Tj,n
T = Tj,n + k P k − 1,n
t  loss,n Tj,kn −θT 
k

 j,n CΔ1,n 1,n1, n
 T k +1
= T k
+  P k
− 


C1, n  T k −T k θ1, n k  k 

j, n j, n loss, n





k + ∆t T − T2,n
T1,n

k +1 j,n 1,n
− 1,n

 = T1,n
Δt  T j , n − T1, n T1, n − T2, n 
C2,n k θ1,n k

k θ2,n k
  k +1


k
T1, n = T1, n + . . . . . . − 


 C2, n  θ1, n θ 2, n  (4)

 
  k 
k k k

k ...... Tm−2,n − Tm−1,n T −T
Tm−1,n = Tm−1,n
k +1
+ C∆t − m−θ1,nm,n c,n (4)




 m,n θm−1,n

T k +1 = T k + Δt k Tm − 2, n − 
k k k k
Tm −1, n Tm − 1, n − Tc , n



k −− T k 

k
 Tm−1,n − Tc,n 

m1−1, n m −1, n ∆t Tc,n
Tc,n = Tc,n + C Cm , n  θ θ m −1, n− k θ m,n

k + k e




ce,n m,n θc,n
 Δt  Tm −1, n − Tc, n T!
k k k k

c, n − Te
Tc,kn+1 = Tc,kn +  k −− k k− T k
!
 Tek+1 = Tek + ∆t
Cce, n  Σ θ m ,c,n T T e θ k T e a
 n −
c,n k (5)
Ce n θce,n θe
where ∆t is the discrete step size.
Δt   Tc, n − Te  Tek − Tak 
k k
k +1
Then, the convectiveTheat = T k
+
resistance of the housing − at ktime k is expressed as (5)
Ce  n  θce, n  θe 
e e
 
Te − Ta ∆t
k k
where Δt is the discrete step size.
θek = (6)
Then, the convective heat resistance Ce Tekof+the Gnk − CeatTektime
∆t∑housing +1
k is expressed as
n
Electronics 2024, 13, 1665 9 of 19

where
k − Tk
Tc,n e
Gnk = (7)
θce,n
It can be found from (5) that the convective heat resistance θek of the converter housing
k , housing temperature
can be calculated by using the historical data of case temperature Tc,n
k k
Te and ambient temperature Ta . The extraction of convectivenheat o resistance samples is
completed by forming the calculated results into a time series θek .

3.2. Principle of Wavelet Packet Transform

Since the variation in airflow rate in the natural environment has obvious frequency
domain characteristics, the convective heat resistance of converter housing is a special
random process with frequency characteristics. To grasp the variation rule of convective
heat resistance, it is necessary to analyze the samples of convective heat resistance in a
time–frequency domain.
Wavelet packet transform is a typical time–frequency domain analysis method that
reflects the variation characteristics of signals in different frequency bands by using different
groups of wavelet packet coefficients [24]. The wavelet packet coefficients are obtained by
wavelet decomposition, which is given by

2z,v z,w
dl +1 = ∑ h(w−2v) dl

w
+1,v
(8)
d2z

l +1 = ∑ g(w−2v) dz,w
l
w

where l is the number of decomposition layers, which affects the number of decomposed
frequency bands. z is the wavelet packet coefficient number. v and w represent different
2z+1,v
time steps. d2z,v
l +1 and dl +1 are the even- and odd-numbered wavelet packet coefficients
in the l + 1 layer, respectively. h(w−2v) and g(w−2v) are the low-pass and high-pass filter
coefficients, respectively, which are related to the selected wavelet basis function.
The decomposed signal can be restored by the layer-by-layer reconstruction of wavelet
packet coefficients. Wavelet packet reconstruction is expressed as

dz,v
l = ∑ [eh(v−2w) d2z,w 2z+1,w
l +1 + g(v−2w) dl +1
e ] (9)
v

where eh(v−2w) and ge(v−2w) are the low-pass and high-pass filter coefficients of wavelet
packet reconstruction, respectively.

3.3. Time–Frequency Domain Characteristics of Housing Convective Heat Resistance

The convective heat resistance of the converter housing in the natural air-cooling envi-
ronment shown in Figure 8 is extracted with (5). The extracted convective heat resistance
samples are shown in Figure 6. Then, these samples are analyzed by the wavelet packet
transform. The basis function of the wavelet packet transform is 30 dB wavelets, and the
decomposition layer l is three layers.
After wavelet packet decomposition, the wavelet packet coefficients in eight frequency
bands are obtained. These wavelet packet coefficients are represented by d1,v 8,v
3 to d3 , and
are shown in Figure 9. Since the upload period of experimental data in the experimental
platform is 1 s, the signal frequency range that the wavelet packet transform can analyze is
0 Hz to 0.5 Hz.
 20

Convective thermal resistance(℃/W)

18
Electronics 2024, 13, 1665 10 of 19
Electronics 2024, 13, 1665 16 10 of 19

14

12
20

10

Convective thermal resistance(℃/W)

18

8
16 0 20 40 60 80 100 120 140 160 180
t(min)
14
Figure 8. Convective heat resistance samples of the converter housing.
12

After wavelet packet decomposition, the wavelet packet coefficients in eight fre-
10
quency bands are obtained. These wavelet packet coefficients are represented by d31, v to
d38, v8 ,0 and20are shown
40 60 in Figure
80 100 9. Since
120 the upload
140 160 180 period of experimental data in the ex-
t(min)
perimental platform is 1 s, the signal frequency range that the wavelet packet transform
can analyze
Figure is 0 Hz toheat
8. Convective 0.5 Hz.
resistance samples of the converter housing.
Figure 8. Convective heat resistance samples of the converter housing.

46 3
After wavelet packet decomposition, the wavelet packet coefficients in eight fre-
40
quency bands are obtained. These wavelet packet coefficients are represented by d31, v to
d38, v , and are shown in Figure 9. 0Since the upload period of experimental data in the ex-
d 2,ν
3
d 1,ν
3

32
perimental platform is 1 s, the signal frequency range that the wavelet packet transform
can analyze is 0 Hz to 0.5 Hz. -3
28 0 20 40 60 80 100 120 140 160 180
0 20 40 60 80 100 120 140 160 180
t(min)
t(min)
46 3
(a) (b)
40
3 3
d 2,ν

0
3
d 1,ν
3

32
0 0
d 33,ν

d 4,ν

-3
3

28 0 20 40 60 80 100 120 140 160 180

0 20 40 60 80 100 120 140 160 180
t(min)
t(min)

-3
0 20 40 60 (a)
80 100 120 140 160 180
-3
0 20 40 60 80 100
(b) 120 140 160 180
t(min) t(min)
3 3
(c) (d)
3 3

0 0
d 33,ν

d 4,ν
3
d 36,ν
d 5,ν

0 0
3

-3 -3
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
t(min) t(min)
-3
-3
0 20 40 60 80 100 120 140 160 180
0 20 40 60 80 100 120 140 160 180
t(min)
(c) t(min) (d)
3 (e) 3 (f)

Figure 9. Cont.
d 36,ν
d 5,ν

0 0
3

-3
-3
0 20 40 60 80 100 120 140 160 180
0 20 40 60 80 100 120 140 160 180
t(min) t(min)

(e) (f)
Electronics 2024, 13, 1665 11 of 19
Electronics 2024, 13, 1665 11 of 19

3 3

d 8,ν
0
d 7,ν

3
3

-3 -3
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
t(min) t(min)

(g) (h)

Figure
Figure 9.
9. Wavelet packetcoefficients
Wavelet packet coefficientsinindifferent
different frequency
frequency bands.
bands. (a) (a) Frequency
Frequency bandband
1 (0 1to(00.0625
to 0.0625
Hz),
Hz), (b) frequency band 2 (0.0625 to 0.125 Hz), (c) frequency band 3 (0.125 to 0.1875
(b) frequency band 2 (0.0625 to 0.125 Hz), (c) frequency band 3 (0.125 to 0.1875 Hz), (d) frequencyHz), (d) fre-
quency band 4 (0.1875 to 0.25 Hz), (e) frequency band 5 (0.25 to 0.3125 Hz), (f) frequency band 6
band 4 (0.1875 to 0.25 Hz), (e) frequency band 5 (0.25 to 0.3125 Hz), (f) frequency band 6 (0.3125 to
(0.3125 to 0.375 Hz), (g) frequency band 7 (0.375 to 0.4375 Hz), (h) frequency band 8 (0.4375 to 0.5
0.375 Hz), (g) frequency band 7 (0.375 to 0.4375 Hz), (h) frequency band 8 (0.4375 to 0.5 Hz).
Hz).
The variations in wavelet packet coefficients directly reflect the fluctuation degree of
The variations in wavelet packet coefficients directly reflect the fluctuation degree of
the signals in various frequency bands. It is easy to find that the wavelet packet coefficients
the
havesignals
a largeinamplitude
various frequency bands.
change in the It is easy
frequency bandto of
find that
0 Hz to the wavelet
0.0625 packet
Hz, which coeffi-
indicates
cients have a large amplitude change in the frequency band of 0 Hz to 0.0625
that the fluctuation in convective heat resistance in the low-frequency band is more intense Hz, which
indicates that the fluctuation in convective heat resistance in the low-frequency
than that in the high-frequency band. The variation law of convective heat resistance affects band is
more intense than that in the high-frequency band. The variation law of convective
the fluctuation characteristics of the device junction temperature, so it should be considered heat
resistance affects
in the thermal the assessment.
safety fluctuation characteristics of the device junction temperature, so it
should be considered in the thermal safety assessment.
4. The Proposed Thermal Safety Assessment Method
4. TheInProposed Thermal
this section, a methodSafety Assessment
is proposed Method
to assess the thermal safety of natural air-cooled
In this section,
converters a method
in the form of theisover-temperature
proposed to assessprobability
the thermalofsafety
power of devices.
natural air-cooled
Since the
converters in the
thermal safety form of the
assessment over-temperature
involves how to reflectprobability of power devices.
random fluctuations in deviceSince the
junction
thermal safetythe
temperature, assessment involves
Monte Carlo how to
principle forreflect
randomrandom fluctuations
simulation is first in device junction
introduced. Then,
temperature, thethe
the principle of Monte Carlomethod
proposed principle
andfortherandom simulation risk
over-temperature is first introduced.
function Then,
of the device
the
are principle of thethe
given. Finally, proposed method
proposed andisthe
method over-temperature
verified risk function
with the experimental of the device
data.
are given. Finally, the proposed method is verified with the experimental data.
4.1. Monte Carlo Principles
4.1. Monte Carloidea
The basic Principles
of the Monte Carlo method is to transform the problem to be solved into
a probability
The basic idea of the calculation
distribution Monte Carlo bymethod
randomissampling
to transformand then obtain the
the problem to numerical
be solved
solution of the problem [25]. The calculation process of the Monte
into a probability distribution calculation by random sampling and then obtain the Carlo method mainly
nu-
includes three steps: constructing the probability process, realizing probability
merical solution of the problem [25]. The calculation process of the Monte Carlo method distribution
sampling,
mainly and establishing
includes an estimatorthe
three steps: constructing [26]. Firstly, process,
probability the problem should
realizing be correctly
probability dis-
described as a probability process, as some variables in this process
tribution sampling, and establishing an estimator [26]. Firstly, the problem shouldare exactly the solution
be cor-
to the described
rectly problem. Then, the randomprocess,
as a probability samplingas process of the Monte
some variables in thisCarlo method
process is realized
are exactly the
solution to the problem. Then, the random sampling process of the Monte Carlosampling
by using random variables with known probability distributions. Finally, the method
results
is are calculated
realized and counted
by using random so aswith
variables to obtain
known an probability
unbiased estimation of theFinally,
distributions. solutiontheto
the problem.
sampling results are calculated and counted so as to obtain an unbiased estimation of the
In the Monte Carlo method, the arithmetic mean of the simple sample Z1 , Z2 . . . Zn of
solution to the problem.
the random variable Z is used as the approximate solution to the problem, as shown in the
In the Monte Carlo method, the arithmetic mean of the simple sample Z1, Z2 … Zn of
following formula:
the random variable Z is used as the approximate solution to the problem, as shown in
1 N
the following formula: ZN = ∑ Zi (10)
N i =1
1 N
where N is the total number of samples and Z N Z=N is theZi arithmetic mean value calculated from (10)
N i =1
N random variable samples. According to the law of large numbers, if Z1 , Z2 . . . Zn satisfies
where N is the total number of samples and ZN is the arithmetic mean value calculated
from N random variable samples. According to the law of large numbers, if Z1, Z2 … Zn
satisfies the independent and identical distribution and has a finite expected value (E(Z)
< ∞), then the formula can be obtained.
Electronics 2024, 13, 1665 12 of 19

Electronics 2024, 13, 1665 12 of 19

the independent and identical distribution and has a finite expected value (E(Z) < ∞), then
the formula can be obtained.

 1N N 
!
1 ⎯⎯⎯ → E(Z ) =
P P  ∑ZZ i i−−−N→
→∞ E ( Z )  =11 (11)
N iN
=1
i =1 N →∞ 
That is,
That is, when
whenthe thenumber
numberofofsamples
samplesofofrandom
random variable
variable ZZ is is large
large enough,
enough, thethe aver-
average
age value of all sub-samples will converge to the expected value of the solution
value of all sub-samples will converge to the expected value of the solution to the problem to the
problem with a probability
with a probability of 1. of 1.
In this
In this paper,
paper,thethewavelet
waveletpacket
packetcoefficients
coefficientsextracted
extractedfrom
from thethe convective
convective heat
heat re-
resis-
sistance
tance samples
samples are statistically
are statistically analyzed
analyzed by thisbymethod,
this method,
and theand the probability
probability density
density function
function
is is obtained.
obtained. Finally, theFinally, the convective
convective heatcomponents
heat resistance resistance components in different
in different frequency fre-
bands
quency bands
are calculated. are calculated.

4.2. Proposed Assessment Method

Based on wavelet packet transform and the Monte Carlo principle, a thermal safety
assessment method for power devices in natural
natural air-cooled
air-cooled converters
converters is proposed,
proposed, as
shown in Figure
Figure 10.
10. The proposed
proposed method
method consists
consists of
of four
four main
main steps.
steps.

convective thermal resistance

sample extraction
convective thermal resistance
sample sequence θ e { }
k

Step1: time-frequency
domain analysis
wavelet packet decomposition

wavelet packet wavelet packet wavelet packet

1,v 2,v … 3,v
coefficient d l coefficient dl coefficient dl

probability statistics

probability density probability density probability density

…
function f1 function f 2 function f z

random simulation and wavelet

packet reconstruction Step 2: random simulation of
junction temperature
simulated sequences of k
different converter
{ }
convective thermal resistance θ e
current
thermal network model
different ambient
temperature simulated sequences of
{ }
junction temperature Tj, n
k

probability statistics
Step 3: over-temperature
probability calculation
over-temperature probability Pr , n

curve fitting
specific converter
current
over-temperature probability Step 4: thermal safety
function Pr , n ( I , Ta ) assessment
specific ambient
temperature
thermal safety assessment results

Figure 10. Principle block diagram of the proposed method.

Figure 10. Principle block diagram of the proposed method.

{ }
n o
Step
Step 11isisapplied
appliedtotoextract
extractthe convective
the convectiveheat resistance
heat samples
resistance samplesθek θofek theofconverter
the con-
housing and perform
verter housing the time–frequency
and perform domain
the time–frequency analysis,
domain as described
analysis, in theinprevious
as described the pre-
section. Step 2 Step
vious section. is mainly used toused
2 is mainly realize
to the Monte
realize theCarlo
Montesimulation of the device
Carlo simulation of thejunction
device
junction temperature. For this purpose, the wavelet packet coefficients dlz , v of convective
heat resistance in the different frequency bands are statistically analyzed to construct their
Electronics 2024, 13, 1665 13 of 19

temperature. For this purpose, the wavelet packet coefficients dz,v l of convective heat resis-
tance in the different frequency bands are statistically analyzed to construct their probability
density functions (PDFs) f 1 to f z . According to these PDFs, the wavelet packet coefficients
n o
are randomly simulated and reconstructed to generate several simulated sequences θeek
n o
for convective heat resistance. The simulated sequences T ek of device junction temper-
n oj
ature can be generated by substituting the sequence θeek , different converter current I
and ambient temperature Ta into the equivalent thermal network model of the converter.
In step 3, the over-temperature probability Pr,n of the device n, which is related to the
converter current and ambient temperature, can be obtained with the probability statistics
of the simulated junction temperature. In the last step, an over-temperature probability
function Pr,n ( I, Ta ) is established by data fitting and used to assess the thermal safety of the
device at the specific converter current and ambient temperature.
Since the wavelet packet transform and Monte Carlo methods describe the random
effect of convective heat resistance on the device junction temperature under different
frequency bands, the junction temperature fluctuation can be effectively simulated in the
natural air-cooling environment. With the junction temperature obtained by random simu-
lation, the proposed method is able to assess the thermal safety of the device in the form of
over-temperature probability under different converter currents and ambient temperatures.

4.3. Over-Temperature Risk Function

It is well known that under specific thermal convection conditions, the over-temperature
risk of power devices increases with the increase in power loss and ambient temperature.
The converter used in this study is a voltage source converter with output square-wave
voltage. Since the amplitude of the square-wave output voltage and the bus voltage is
constant, the device loss (including switching loss and on-state loss) in the converter is
positively correlated with the load current I. Because of this, a two-variable growth function
related to the converter load current and ambient temperature is employed to describe the
over-temperature probability of the device. The growth function used in this paper is a
Pearl function, which has the advantages of a simple structure and fewer variables [27].
According to the principle of Pearl function, the over-temperature probability Pr,n (I, Ta ) of
device n can be expressed as

1
Pr,n ( I, Ta ) = (12)
1 + α( Ta ) · e− β(Ta )( I −γ(Ta ))

where α(Ta ), β(Ta ) and γ(Ta ) are coefficients of the Pearl function with respect to the
ambient temperature Ta .
These coefficients can be obtained by fitting polynomials, which are given by

α( Ta ) = ∑ µi Tai (13)
i =0

β( Ta ) = ∑ δi Tai (14)
i =1

γ( Ta ) = ∑ ε i Tai (15)
i =1

where µi , δi and εi are polynomial coefficients.

If the amplitude of the output voltage or the bus voltage is dynamically changing, the
load current cannot accurately reflect the device loss. Consequently, for some converters
such as current source inverters or battery chargers, the converter power should be used
instead of the load current as the variable to calculate the over-temperature probability of
the power device.
Electronics 2024,
Electronics 13, 13,
2024, 16651665 14 of 19
14 of 19

4.4.
4.4.Validation
Validation and
and Analysis
Analysis
Theprobability
The probability of
of the
the wavelet
waveletpacket
packetcoefficients
coefficientsforfor
thethe
convective
convectiveheatheat
resistance
resistance
samples shown in Figure 7 is statistically analyzed by the proposed
samples shown in Figure 7 is statistically analyzed by the proposed method. method. Figure 11 11
Figure
shows the probability density function of the wavelet packet coefficients in different
shows the probability density function of the wavelet packet coefficients in different fre- fre-
quencybands.
quency bands. From
From the
the results,
results,ititcan
canbebefound
foundthat thethe
that wavelet packet
wavelet coefficients
packet in thein the
coefficients
experiment in this paper basically obey the normal distribution.
experiment in this paper basically obey the normal distribution.

μ = 37.19059

Probability density
0.20 μ = 3.85473E-4
Probability density

σ = 2.38248 1.0 σ = 0.39781

0.15

0.10
0.5
0.05

0.00 0.0
28 32 36 40 44 -1.6 -0.8 0.0 0.8 1.6
wavelet packet coefficient wavelet packet coefficient
(a) (b)

Probability density
μ = -1.24857E-4
Probability density

1.0 μ = 4.83168E-5
1.0 σ = 0.34668 σ = 0.41667

0.5
0.5

0.0 0.0
-1.6 -0.8 0.0 0.8 1.6 -0.8 0.0 0.8 1.6
wavelet packet coefficient wavelet packet coefficient

(c) (d)

1.0
Probability density

Probability density

μ = 4.07324E-5 μ = -3.24537E-5
σ = 0.49988 1.0 σ = 0.43463

0.5
0.5

0.0 0.0
-2 0 2 -1.6 -0.8 0.0 0.8 1.6
wavelet packet coefficient wavelet packet coefficient
(e) (f)

1.0
Probability density

1.0 μ = 1.13009E-4
Probability density

μ = -1.46317E-4
σ = 0.48102 σ = 0.41876

0.5 0.5

0.0 0.0
-2 -1 0 1 2 -0.8 0.0 0.8 1.6
wavelet packet coefficient wavelet packet coefficient
(g) (h)

Figure 11. The probability density functions for the wavelet packet coefficients of convective heat
Figure 11. The probability density functions for the wavelet packet coefficients of convective heat
resistance under different frequency bands. (a) Probability density function f1, (b) probability
resistance under different frequency bands. (a) Probability density function f 1 , (b) probability density
function f 2 , (c) probability density function f 3 , (d) probability density function f 4 , (e) probability den-
sity function f 5 , (f) probability density function f 6 , (g) probability density function f 7 , (h) probability
density function f 8 .
Electronics 2024, 13, 1665 15 of 19

density function f2, (c) probability density function f3, (d) probability density function f4, (e) proba-
Electronics 2024, 13, 1665 bility density function f5, (f) probability density function f6, (g) probability density function f7, (h)
15 of 19
probability density function f8.

The simulation sequences of convective heat resistance are obtained by the random
The simulation sequences of convective heat resistance are obtained by the random
simulation and reconstruction of the above wavelet packet coefficients. The number of
simulation and reconstruction of the above wavelet packet coefficients. The number of
convective heat resistance sequences simulated in this paper is 200. The greater the num-
convective heat resistance sequences simulated in this paper is 200. The greater the number
ber of simulation sequences, the richer the simulation samples. This is beneficial to im-
of simulation sequences, the richer the simulation samples. This is beneficial to improve
prove the statistical accuracy of over-temperature probability, but it will increase the
the statistical accuracy of over-temperature probability, but it will increase the amount of
amount of calculation. Therefore, it is necessary to reasonably select the number of simu-
calculation. Therefore, it is necessary to reasonably select the number of simulations in
lations
practicalin practical
engineeringengineering
accordingaccording
to hardware to hardware
resources.resources.
Figure
Figure12 12shows
showsaacomparison
comparisonof ofthe
thepower
powerdensity
densityspectrum
spectrumbetween
betweenthe
thesimulated
simulated
sequence and the sample sequence of convective heat resistance. Power
sequence and the sample sequence of convective heat resistance. Power spectral density spectral density
is a
israndom
a randomprocess analysis method that reflects the distribution characteristics of signalsignal
process analysis method that reflects the distribution characteristics of power
power at different
at different frequencies
frequencies [28].
[28]. The The similarity
similarity between between
differentdifferent power spectral
power spectral den-
densities can
sities can be used to analyze the difference in random characteristics
be used to analyze the difference in random characteristics between different random between different
random
processes. processes.
According According to the Pearson
to the Pearson correlation
correlation coefficient
coefficient calculation
calculation method method
[29], the
[29], the similarity of the power spectral density between the simulated
similarity of the power spectral density between the simulated sequence and the sample sequence and the
sample sequence is 98.53%. Since the similarity is close to 1, the random
sequence is 98.53%. Since the similarity is close to 1, the random characteristics between characteristics
between the simulated
the simulated sequence sequence and the
and the sample sampleare
sequence sequence
basicallyare
thebasically theindicates
same. This same. This
that
indicates that the proposed method accurately simulates the random
the proposed method accurately simulates the random characteristics of convective heat characteristics of
convective
resistance in heat resistance
natural in natural
air-cooling air-cooling environments.
environments.

Figure
Figure12.
12.Power
Powerspectral
spectraldensities
densitiesof
ofthe
thesample
samplesequence
sequenceand
andthe
thesimulated
simulatedsequences.
sequences.

Next,the
Next, thesimulated
simulatedsequences
sequencesof ofconvective
convectiveheat heatresistance
resistancearearesubstituted
substitutedintointothe
the
thermal network model to simulate the random variations in device
thermal network model to simulate the random variations in device junction temperature junction tempera-
atture at different
different currents
currents and ambient
and ambient temperatures.
temperatures. FigureFigure
13 shows13 shows fiveofsets
five sets of random
random sim-
simulation
ulation results
results forjunction
for the the junction temperature
temperature of device
of device Q1simulated
Q1. All . All simulated sequences
sequences of
of con-
convective
vective heatheat resistance
resistance are are used
used to simulate
to simulate thethe junction
junction temperaturechange.
temperature change.Therefore,
Therefore,
200groups
200 groupsof ofjunction
junctiontemperature
temperaturesimulation
simulationresults
resultscan
canbe beobtained
obtainedunder
undereach
eachcombi-
combi-
nation of ambient temperature and current. However, for the convenience
nation of ambient temperature and current. However, for the convenience of observation, of observation,
onlyfive
only fivegroups
groupsofofjunction
junctiontemperature
temperaturesimulation
simulationresults
resultsare
arepresented
presentedhere.
here.Obviously,
Obviously,
the simulated junction temperatures are always in random
the simulated junction temperatures are always in random fluctuation. fluctuation.
The over-temperature probability of the device Q1 under different converter currents
and ambient temperatures is obtained with the simulated junction temperatures. By using
fifth-order polynomial fitting, the Pearl function for calculating the probability of over-
temperature can be established. The fitting coefficients are listed in Table 2 and Figure 14,
showing the over-temperature probability surface calculated with the Pearl function. With
the good computing power of the microprocessor in this study, the above-mentioned
random simulation of junction temperature and the establishment of the over-temperature
probability are achieved in 200 ms.
 Ta=30℃，I=7.5A

Junction temperature
80 Ta=25℃，I=6.5A
Ta=20℃，I=5.5A
60
Ta=15℃，I=4.5A
40
Electronics 2024,
Electronics 13,13,
2024, 1665
1665 1616 of 19
of 19
20

0 60 120 180
t(min)

Figure
120
13. Power spectral densities of the sample sequence and the simulated sequences.
Ta=35℃，I=8.5A

Junction temperature(℃)
100Ta=30℃，I=7.5A of the device Q1 under different converter currents
The over-temperature probability
and80ambient temperatures is obtained with the simulated junction temperatures. By using
Ta=25℃，I=6.5A

fifth-order polynomial Ta=20℃，I=5.5A

fitting, the Pearl function for calculating the probability of over-
60
Ta=15℃，I=4.5A
temperature can be established. The fitting coefficients are listed in Table 2 and Figure 14,
40
showing the over-temperature probability surface calculated with the Pearl function. With
the good
20 computing power of the microprocessor in this study, the above-mentioned ran-
dom simulation
0 of 60junction temperature
120 and the
180 establishment of the over-temperature
t(min)
probability are achieved in 200 ms.
Figure
Figure13.
13.Power
Powerspectral
spectral densities of the
densities of thesample
samplesequence
sequenceand
andthe
the simulated
simulated sequences.
sequences.
Table 2. Coefficients of the Pearl growth function.
Table 2. Coefficients
The of the Pearl
over-temperature growth function.
probability of the device Q1 under different converter currents
Fitting Coefficient Value
and ambient temperatures μ1
is obtained with the simulated junction
−2449.27
temperatures. By using
Fitting Coefficient Value
fifth-order polynomial μ2 fitting, the Pearl function for calculating the probability of over-
µ1 −976.43
2449.27
temperature can be established.
μµ3 The fitting coefficients are listed in
−73.51 Table 2 and Figure 14,
2 976.43
showing the over-temperature
μµ43 probability surface calculated with−2.04 the Pearl function. With
73.51
the good computing power μµ54 of the microprocessor in this study, the
2.04 above-mentioned ran-
−0.019
dom simulation of junctionδµ1 5 temperature and the establishment −6.88
0.019
of the over-temperature
δ δ21 6.88
−0.11
probability are achieved in 200 ms.
δδ32 −0.11
−0.0039
δ δ43 −0.0039
0.00033
Table 2. Coefficients of theδ Pearl growth function. 0.00033
δ54 0.0000055
δ5 0.0000055
Fitting Coefficient
ε1 Value
3.80
ε1 3.80
εμε21 0.1816
−2449.27
0.1816
2
εμε332 −0.01567
976.43
−0.01567
εμε443 0.000547
0.000547
−73.51
εμε554 0.0000062
0.0000062
2.04
μ5 −0.019
δ1 6.88
δ2 −0.11
δ3 −0.0039
δ4 0.00033
δ5 0.0000055
ε1 3.80
ε2 0.1816
ε3 −0.01567
ε4 0.000547
ε5 0.0000062

Figure14.
Figure Over-temperatureprobability
14.Over-temperature probability surface
surface of
of the
the device
device Q
Q11 calculated
calculated with
with the
the Pearl
Pearl func-
function.
tion.
Finally, the over-temperature risks of the device Q1 under five groups of specific
converter current and ambient temperature, as shown in Table 3, are estimated by using the
above Pearl function. The data listed in this table show that the risk of over-temperature in
groups #1 and #2 reaches more than 90%, which also indicates that the converter has a great
probability of overheating failure. The over-temperature risk in group #3 is 3.31%, which
means that the converter has relatively good safety, but there is still a small probability of
thermal failure. In groups #4 and #5, the over-temperature risk of the device is 0, indicating
that the thermal safety of the converter is extremely high.

Figure 14. Over-temperature probability surface of the device Q1 calculated with the Pearl func-
tion.
 in groups #1 and #2 reaches more than 90%, which also indicates that the converter has a
great probability of overheating failure. The over-temperature risk in group #3 is 3.31%,
which means that the converter has relatively good safety, but there is still a small proba-
bility of thermal failure. In groups #4 and #5, the over-temperature risk of the device is 0,
Electronics 2024, 13, 1665 indicating that the thermal safety of the converter is extremely high. 17 of 19

Table 3. Over-temperature risk assessment results of device Q1 at specific converter currents and
ambient
Table 3.temperatures.
Over-temperature risk assessment results of device Q1 at specific converter currents and
ambient temperatures.
Converter Current Ambient Temperature Ta Over-Temperature Proba-
Group
I(A) (°C) Ambient bility (%)
Over-Temperature
Group Converter Current I(A) ◦
#1 8.9 35
Temperature Ta ( C) 94.58
Probability (%)
#2 #1 7.9 8.9 30 35 90.2694.58
#3 #2 6.9 7.9 25 30 3.3190.26
#4 #3 5.9 6.9 20 25 0 3.31
#4 5.9 20 0
#5 4.9 15 0
#5 4.9 15 0

To verify the above thermal safety assessment results, the converter in the experiment
To verify the above thermal safety assessment results, the converter in the experiment
platform is operated under the current and ambient temperatures shown in Table 3 and
platform is operated under the current and ambient temperatures shown in Table 3 and
the natural air-cooled environment shown in Figure 4. The junction temperature variation
the natural air-cooled environment shown in Figure 4. The junction temperature variation
in
indevice
deviceQQ 1 in the experiment is shown in Figure 15. It can be seen that the junction tem-
1 in the experiment is shown in Figure 15. It can be seen that the junction
perature
temperatureexceeds Tjmax Tin groups
exceeds #1 and #2. The junction temperature in group #3 is less
jmax in groups #1 and #2. The junction temperature in group #3 is
than Tjmax, but its maximum value is close to Tjmax. In groups #4 and #5, the junction tem-
less than Tjmax , but its maximum value is close to Tjmax . In groups #4 and #5, the junction
perature
temperatureis much smaller
is much than than
smaller Tjmax.TThese experimental phenomena are consistent with
jmax . These experimental phenomena are consistent
the results of the thermal safety assessment.
with the results of the thermal safety assessment. This indicates
This indicatesthat, based
that, basedononthethewavelet
wavelet
packet transform and Monte Carlo methods, the proposed method
packet transform and Monte Carlo methods, the proposed method effectively simulateseffectively simulates
the
the influence
influence of random
of random airflow
airflow on on
thethe device
device junction
junction temperature
temperature of converters
of converters under
under the
the natural
natural air-cooled
air-cooled environments.
environments. TheThe thermal
thermal safety
safety of other
of other devices
devices in the
in the converter
converter can
can also be evaluated by the proposed method, but this is not
also be evaluated by the proposed method, but this is not explored here. explored here.

140 Ta=35℃，I=8.9A
Junction temperature(℃)

120 Ta=30℃，I=7.9A

100 Ta=25℃，I=6.9A
Ta=20℃，I=5.9A
80
Ta=15℃，I=4.9A
60

40

20

0 60 120 180
t(min)

Figure
Figure15.
15.Junction
Junctiontemperature
temperaturevariation
variationin
indevice
deviceQQ11ininthe
thevalidation
validationexperiment
experimentofofthe
theproposed
proposed
method.
method.

The
Theabove
aboveresults
resultsshow
showthetheeffectiveness
effectivenessofofthe
theproposed
proposedmethod
methodforforassessment
assessmentof of
the thermal safety
the safetyofofpower
power devices
devicesin natural air-cooled
in natural converters.
air-cooled The proposed
converters. method
The proposed
can be used
method inused
can be the test stage
in the testorstage
practical application
or practical stage of
application converter
stage products.
of converter In the
products.
test
In thestage, the convective
test stage, heat heat
the convective resistance sample
resistance of the
sample of housing is extracted
the housing through
is extracted the
through
off-line
the experiment
off-line experiment of prototypes,
of prototypes,andandthethe
heat-dissipation
heat-dissipationperformance
performanceof the prototype
of the proto-
is improved
type according
is improved to the
according to estimated
the estimatedover-temperature
over-temperature probability of devices.
probability In the
of devices. In
practical application stage, the convective heat resistance samples should
the practical application stage, the convective heat resistance samples should be be continuously
updated and then the proposed method can estimate and predict the over-temperature risk
of devices in real time.

5. Conclusions
In this paper, a thermal safety assessment method for power devices in natural air-
cooled converters is proposed. The experimental results show that by using wavelet packet
transform and Monte Carlo simulation, the proposed method can effectively simulate the
influence of random airflow in natural air-cooling environments on device junction temper-
ature in converters. In this way, the over-temperature risk of power devices under different
ambient temperature and current levels can be accurately estimated in the form of proba-
bility, which effectively improves the thermal reliability of natural air-cooled converters.
Electronics 2024, 13, 1665 18 of 19

It should be noted that the method proposed in this study is only for natural air-cooled
converters with good thermal conductivity and small size. If the size of the converter
housing is large or the thermal conductivity of the housing material is low, the proposed
method needs to be further improved due to the non-negligible surface temperature
distribution on the housing. The time–frequency domain analysis of convective heat
resistance and the random simulation of device junction temperature in this paper have
provided a good foundation for the future improvement of this method. In addition,
to improve the reliability of junction temperature estimation, more accurate and novel
thermal networks can be used in future to replace the linear thermal network used in the
proposed method.

Author Contributions: Conceptualization, Y.Z.; Investigation, W.H.; Methodology, W.H., Y.Z. and
Z.Y.; Writing—original draft, W.H.; Writing—review and editing, Z.L., J.L., F.B. and W.Z. All authors
have read and agreed to the published version of the manuscript.
Funding: This research was funded by the National Natural Science Foundation of China (Grant No.
42206225) and the Project of State Key Laboratory of Smart Grid Protection and Operation Control
(SGNR0000KJJS2007613).
Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. Qian, C.; Gheitaghy, A.M.; Fan, J.; Tang, H.; Sun, B.; Ye, H.; Zhang, G. Thermal management on IGBT power electronic devices
and modules. IEEE Access 2018, 6, 12868–12884. [CrossRef]
2. Bahman, A.S.; Ma, K.; Blaabjerg, F. A lumped thermal model including thermal coupling and thermal boundary conditions for
high-power IGBT modules. IEEE Trans. Power Electron. 2018, 33, 2518–2530. [CrossRef]
3. Chen, H.; Yang, J.; Xu, S. Electrothermal-based junction temperature estimation model for converter of switched reluctance motor
drive system. IEEE Trans. Ind. Electron. 2020, 67, 874–883. [CrossRef]
4. Lutz, J.; Schlangenotto, H.; Scheuermann, U.; Doncker, R.D. Semiconductor Power Devices; Springer: Berlin, Germany, 2011;
pp. 360–368.
5. Drofenik, U.; Cottet, D.; Müsing, A.; Meyer, J.M.; Kolar, J.W. Modelling the thermal coupling between internal power semiconduc-
tor dies of a water-cooled 3300V/1200A HiPak IGBT module. In Proceedings of the International Exhibition and Conference for
Power Electronics, Intelligent Motion and Power Quality (PCIM Europe 2007), Nuremberg, Germany, 22–24 May 2007.
6. Wu, R.; Iannuzzo, F.; Wang, H.; Blaabjerg, F. Fast and accurate Icepak-PSpice co-simulation of IGBTs under short-circuit with an
advanced PSpice model. In Proceedings of the 7th IET International Conference on Power Electronics, Machines and Drives
(PEMD 2014), Manchester, UK, 8–10 April 2014; pp. 1–5.
7. Greco, G.; Vinci, G.; Bazzano, G.; Raciti, A.; Cristaldi, D. Layered electro-thermal model of high-end integrated power electronics
modules with IGBTs. In Proceedings of the IECON 2014-40th Annual Conference of the IEEE Industrial Electronics Society, Dallas,
TX, USA, 29 October–1 November 2014; pp. 1575–1580.
8. Lemmens, J.; Vanassche, P.; Driesen, J. Optimal control of traction motor drives under electrothermal constraints. IEEE J. Emerg.
Sel. Top. Power Electron. 2014, 2, 249–263. [CrossRef]
9. Chen, H.; Ji, B.; Pickert, V.; Cao, W. Real-time temperature estimation for power MOSFETs considering thermal aging effects.
IEEE Trans. Device Mater. Reliab. 2014, 14, 220–228. [CrossRef]
10. Du, X.; Li, G.; Li, T.; Sun, P.; Zhou, L. Multi-time scale lifetime evaluation of IGBT modules in the wind power converter. Proc.
CSEE 2015, 35, 6152–6161.
11. Górecki, K.; Górecki, P. Nonlinear Compact Thermal Model of the IGBT Dedicated to SPICE. IEEE Trans. Power Electron. 2020, 35,
13420–13428. [CrossRef]
12. Górecki, K.; Zarebski, J.; Bisewski, D.; Dabrowski,
˛ J. Nonlinear Compact Thermal Model of Power Semiconductor Devices. IEEE
Trans. Compon. Packag. Technol. 2010, 33, 643–647. [CrossRef]
13. Heng, K.; Yang, X.; Wu, X.; Ye, J.; Liu, G. A Temperature-Dependent Physical Thermal Network Model including Thermal
Boundary Conditions for SiC MOSFET Module. IEEE Trans. Electron Devices 2022, 69, 4444–4452. [CrossRef]
14. Górecki, P.; Wojciechowski, D. Accurate Computation of IGBT Junction Temperature in PLECS. IEEE Trans. Electron Devices 2020,
67, 2865–2871. [CrossRef]
15. Yang, X.; Xu, S.; Heng, K.; Wu, X. Distributed Thermal Modeling for Power Devices and Modules with Equivalent Heat Flow
Path Extraction. IEEE J. Emerg. Sel. Top. Power Electron. 2023, 11, 5863–5876. [CrossRef]
Electronics 2024, 13, 1665 19 of 19

16. Zhang, X.; Wang, S.; Xin, L.; Wu, K.; Li, J. A 3-D thermal network model for the temperature monitoring of thermal grease as
interface material. Microelectron. Reliab. 2020, 114, 113816. [CrossRef]
17. Wang, H.; Zhou, Z.; Xu, Z.; Ge, X.; Yang, Y.; Zhang, Y.; Yao, B.; Xie, D. A Thermal Network Model for Multichip Power Modules
Enabling to Characterize the Thermal Coupling Effects. IEEE Trans. Power Electron. 2024, 39, 6225–6245. [CrossRef]
18. Shi, Y.; Liu, J.; Ai, Y.; Chen, S.; Bai, Y. Dynamic IGBT Three-Dimensional Thermal Network Model Considering Base Solder
Degradation and Thermal Coupling between IGBT Chips. IEEE Trans. Transp. Electrif. 2023, 9, 2994–3011. [CrossRef]
19. Ma, M.; Guo, W.; Yan, X.; Yang, S.; Zhang, X.; Chen, W.; Cai, G. A Three-Dimensional Boundary-Dependent Compact Thermal
Network Model for IGBT Modules in New Energy Vehicles. IEEE Trans. Ind. Electron. 2021, 68, 5248–5258. [CrossRef]
20. Long, Y.; Song, Z. Thermal Analysis of High Torque Density Axial-Radial Flux Permanent Magnet Machine Based on Lumped
Parameter Thermal Network. In Proceedings of the 2023 26th International Conference on Electrical Machines and Systems
(ICEMS), Zhuhai, China, 5–8 November 2023; pp. 1443–1448.
21. Zhao, H.; Zhang, X.; Li, J.; Zhang, F.; Zhao, W.; Zhang, Y.; Gerada, D.; Zhang, H. Effectiveness of Thermal Interface Materials on
Electrical Machines Thermal Performance with Heat Pipes. IEEE Trans. Transp. Electrif. 2023, 10, 1274–1285. [CrossRef]
22. Wang, R.; Fan, X.; Li, D.; Qu, R.; Li, L.; Zou, T. Convective Heat Transfer Characteristics on End-Winding of Stator Immersed
Oil-Cooled Electrical Machines for Aerospace Applications. IEEE Trans. Transp. Electrif. 2022, 8, 4265–4278. [CrossRef]
23. Górecki, P. Compact Thermal Modeling of Power Semiconductor Devices with the Influence of Atmospheric Pressure. Energies
2022, 15, 3565. [CrossRef]
24. BURRUS. Introduction to Wavelets and Wavelet Transforms: A Primer; China Machine Press: Beijing, China, 2005.
25. Metropolis, N.; Ulam, S. The Monte Carlo method. J. Am. Stat. Assoc. 1949, 44, 335–341. [CrossRef]
26. Hammersley, J. Monte Carlo Methods; Springer Science & Business Media: Berlin/Heidelberg, Germany, 2013.
27. Moriguchi, K. An approach for deriving growth equations for quantities exhibiting cumulative growth based on stochastic
interpretation. Phys. A Stat. Mech. Its Appl. 2018, 490, 1150–1163. [CrossRef]
28. Christensen, J.H.; Mortensen, J.; Hansen, A.B.; Andersen, O. Chromatographic preprocessing of GC-MS data for analysis of
complex chemical mixtures. J. Chromatogr. A 2005, 1062, 113–123. [CrossRef] [PubMed]
29. Lieskovská, E.; Jakubec, M.; Jarina, R.; Chmulík, M. A review on speech emotion recognition using deep learning and attention
mechanism. Electronics 2021, 10, 1163. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.