Professional Documents
Culture Documents
Automated Current Control Method For Flux-Linkage Measurement of Synchronous Reluctance Machines
Automated Current Control Method For Flux-Linkage Measurement of Synchronous Reluctance Machines
Automated Current Control Method For Flux-Linkage Measurement of Synchronous Reluctance Machines
2, MARCH/APRIL 2020
Abstract—This article presents a novel approach to the dc stand- axis inductance Ld and quadrature axis inductance Lq called the
still measurement of flux linkage and inductance of a synchronous “saliency ratio” [4].
reluctance machine. In the existing standstill test to measure the For an accurate representation and control, and efficiency
flux linkage of a machine, a pulsed voltage is applied to the machine
in open loop. The flux-linkage characteristics are computed using optimization of the SynRM, a model considering the saturation,
the measured response. This article discusses the limitations of the the cross-magnetization, and the rotor position effects is neces-
voltage pulse-based method, and a current control based method is sary [5]–[8]. The flux-linkage characteristics are also useful in
proposed to overcome these limitations. In the proposed method, a maximum torque per ampere control of SynRMs [9]. Accurate
pulsed current in closed loop is applied to the machine at standstill estimate of flux linkage and inductance is required to improve
(shaft locked). The major benefit is that the time response of the
machine can be modified by properly tuning the controller parame- the control performance of a SynRM drive [10]. Investigation
ters such that the number of measurement samples available during on relevance of including saturation and position dependence
the transient is improved. Thus, the measurement process can be in the inductances for accurate dynamic modeling and control
programmed in a real-time processor to automate the measurement of SynRMs showed that model’s magnetic saturation in the
process. The automation leads to bypassing the recording and SynRM model is mandatory to have an accurate prediction for
offline processing of huge datasets. This article also proposes a
method to automatically tune the current controllers required in performance measures, such as output torque, power factor, and
the measurement process. The proposed method is verified by an stability [11].
experiment performed on a 7.5-hp machine. Flux-linkage and inductance measurement methods are evolv-
Index Terms—Cross-magnetization, current control, flux-
ing to meet the needs of machine and drive design. As explained
linkage measurement, inductance measurement, real-time in [12], a parameter-based numerical model of a machine is
measurement, standstill test, synchronous reluctance machine used in the power hardware in the loop emulation of electric
(SynRM). machines. In order to use the experimental data to represent the
machine model, an accurate and quicker measurement method
I. INTRODUCTION is required. There exist several methods to measure the flux
YNCHRONOUS reluctance machines (SynRMs) are linkages/inductances of SynRMs [13]–[21]. Based on the exist-
S becoming increasingly attractive in various applications,
including electric and hybrid electric vehicles [1], owing to their
ing literature, flux-linkage identification methods can be broadly
classified into: rotational tests and standstill tests. Rotational test
low cost, high efficiency, high torque per ampere capability, methods can be further classified into constant speed method
simple and rugged structure, and high operating temperature (CSM) and dynamic testing method (DTM).
capability [2], [3]. The SynRM rotor is laminated pieces of steel The CSM produces highly accurate results [13], [14]. How-
without any cage, magnets, and rotor winding. Its performance ever, the test setup requires a full dynamometer setup, and the
in terms of torque and power factor depends on the ratio of direct test result is an average flux linkage around the rotor positions. In
the DTM presented in [14], the magnetic model of the machine
is identified during acceleration. However, the data acquisition
Manuscript received April 10, 2019; revised July 17, 2019, September 30,
2019, and December 3, 2019; accepted January 13, 2020. Date of publication period become small at higher acceleration torque, and extra
January 19, 2020; date of current version March 17, 2020. Paper 2019-EMC- arrangement may be needed to increase the moment of inertia.
0327.R3, presented at the 2018 IEEE Energy Conversion Congress and Expo- Both the CSM and DTM cannot be used to measure position
sition, Portland, OR, USA, Sep. 23–27, and approved for publication in the
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by the Electric Machines dependent flux linkages.
Committee of the IEEE Industry Applications Society. This work was sup- In standstill tests, an appropriate dc or ac excitation is ap-
ported in part by the Natural Sciences and Engineering Research Council of plied to the machine at standstill. The dc-locked rotor and ac-
Canada/Hydro-Québec Senior Industrial Research Chair entitled “Design and
Performance of Special Electric Machines” held by Prof. P. Pillay at Concordia locked rotor tests are widely established methods to measure the
University and in part by the InnovÉÉ-NSERC CRD under project “Emulation inductances with saturation effects due to currents in the self-
and Design of Electric and Hybrid Electric Vehicle Motor Drive Systems.” winding [15]. However, they cannot be used to measure cross-
(Corresponding author: Pragasen Pillay.)
The authors are with the Department of Electrical and Computer Engineer- magnetization effects and position-dependent inductances. A
ing, Concordia University, Montréal, QC H3G1M8, Canada (e-mail: r_thike@ dc standstill method to determine the variation of machine
encs.concordia.ca; pragasen.pillay@concordia.ca). inductance with the rotor position is discussed in [16]. The
Color versions of one or more of the figures in this article are available online
at https://ieeexplore.ieee.org. authors use a voltage pulse for excitation to measure the induc-
Digital Object Identifier 10.1109/TIA.2020.2967689 tances considering magnetic saturation and inductance variation
0093-9994 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
THIKE AND PILLAY: AUTOMATED CURRENT CONTROL METHOD FOR FLUX-LINKAGE MEASUREMENT OF SynRMs 1465
distortion. In [19] and [20], two voltage sources are used to controllers only once, yet ensure a similar current response
measure the cross-coupling inductances; one for controlling the for all operating points, irrespective of the variation in the
current in one axis, and another for supplying the voltage pulse machine inductance. Such a response is obtained by using a
in another axis. While Yamamoto et al. [19] use the neutral pseudoderivative feedback (PDF) controller with a novel design
connection to supply a phase current, and Kilthau and Pacas [20] technique compared to a PI controller in [31]. The flux linkage
use two extra resistors, both methods are only applicable to is obtained by integrating the measured response. The benefit
measure inductance maps for three-rotor positions (each phase is that the proposed method can be programmed in a real-time
axis). Stumberger et al. [21] and Thike and Pillay [22] propose system to automate the measurement process. The novelty and
usage of a voltage source inverter (VSI) in order to determine contribution of this article are summarized as follows.
the complete flux linkage and inductance map of a permanent 1) The proposed method applies a pulsed current reference
magnet synchronous machine. In methods presented in [21] in closed loop in one axis (d-axis) while controlling the
and [22], the machine rotor is locked at a fixed position, and current in another axis (q-axis) to a fixed value, which is
the VSI applies a voltage pulse in one axis while controlling the not reported in the literature.
current in the other axis without need for dynamometer setup or 2) This article also demonstrates in detail a method to au-
acceleration test. tomatically tune the current controllers required in the
The flux-linkage calculation methods employed in [17]–[22] measurement process.
apply a pulsed voltage to the machine in open loop. These 3) Implementation of the proposed method to compute the
methods require post processing of a huge dataset (several flux linkage and inductance in a real-time system without
gigabytes). In [23] and [24], real-time processing of measured requiring data acquisition and post processing.
data with some approximating functions are presented for stand- In the existing voltage pulse-based standstill tests, the re-
still self-identification of flux characteristics of SynRMs. The sponse time of the machine completely depends on the machine
authors used a high test voltage as a tradeoff between several time constant. For machines with a lower time constant, a
contrasting factors and the sampling time of the system was higher sampling frequency is required for the same computation
100 μs. However, no analysis was presented on the accuracy accuracy as with a lower sampling frequency and higher time
of the measurement due to the limited sampling time. More- constant machine. Due to the closed-loop control in the proposed
over, the presented methods are useful to find the flux-linkage method, the behavior in the response can be controlled to have
characteristics at one rotor position. In real-time emulation sufficiently large samples during the transients, irrespective of
and advanced control techniques, it is desirable to have evenly the machine time constant. Thus, a fixed sampling frequency
spaced data on the flux linkage and inductance variation with system can be used to measure the parameters of a wide range of
the rotor position [25]–[29]. Lee et al. [30] proposed a method machines. This feature makes it applicable for machine parame-
based on the discrete Fourier transform to identify the interior ter identification using a real-time system. The biggest challenge
permanent magnet synchronous machine (IPMSM) flux linkage in the proposed method is the tuning of the controllers without
considering spatial harmonics. Although the proposed identifi- having machine parameters beforehand. This article proposes
cation technique can be implemented in a real-time system, its a method to determine the current controller parameters that
implementation requires an expensive dynamometer test setup ensure the steady state within the defined period irrespective
with precise speed control. In addition, the method requires an of the parameter variation due to saturation effects. The work
accurate estimation of the inverter nonlinear characteristics for presented in this article is an extension of the work presented by
an accurate estimation of machine flux linkages. the same authors in [32].
Hwang et al. in [31] presented a current control based param- This article is organized as follows. Section II presents the
eter identification of synchronous reluctance motor at standstill. details of the machine under test along with the mathemat-
The authors proposed to apply a current pulse in closed loop to ical model of SynRM. A brief review on the voltage pulse
obtain the machine resistance and inductance in two steps. In method (VPM) and its limitations are discussed in Section III.
the first step, the integral gain in the proportional-integral (PI) In Section IV, a current control based flux-linkage measurement
controller is set to zero, and the machine resistance is obtained technique is proposed. It also presents in details an analysis
from the reference current, the measured steady state current, and and design of the controllers needed to implement the proposed
proportional gain. In the next step, the integral gain is computed method. A real-time implementation of the proposed method
using the measured resistance, and the proportional gain is tuned is presented in Section V. Section VI presents the experi-
until a critically damped response is obtained. The machine mental setup and the test results. This article is concluded in
inductance is then obtained as the ratio of proportional gain Section VIII.
and closed-loop bandwidth. Such a method requires tuning the
controller parameters for each operating point, and it cannot
be programmed in a real-time system to automate the machine II. MATHEMATICAL MODEL OF A SYNRM
parameter measurement. Fig. 1(a) shows the geometry of the machine under test. It
This article proposes a novel current control based standstill is a transversely laminated anisotropic rotor-type SynRM. The
method to measure the flux linkage and inductance for Syn- nameplate data and some design details of the test machine are
RMs under locked rotor conditions. In comparison to Hwang given in Table I. Fig. 1(b) and (c) show the q and d axes equivalent
et al. [31], the method presented in this article tune the current circuits of the SynRM in dq model. Iron loss is not considered
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
1466 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 2, MARCH/APRIL 2020
dλq
v q = R s iq + (4)
dt
λd = Ldd (id , iq )id + Ldq (id , iq )iq (5)
λq = Lqd (id , iq )id + Lqq (id , iq )iq . (6)
The machine under test in this article is designed for delta-
connected winding. However, all six terminals of the motor are
available to the user. In order to avoid mathematical operations
that converts line currents to phase currents in delta-connected
machine, the windings are configured for wye connection during
the test. Since the proposed test method is a standstill test with
dc excitation, the applied voltage will still be within the rated
Fig. 1. SynRM under test and its equivalent circuits without core loss com- winding voltage.
ponent. (a) Rotor and stator geometry. (b) q-axis equivalent circuit. (c) d-axis
equivalent circuit.
III. REVIEW: VOLTAGE PULSE-BASED METHOD
For locked rotor conditions, the dynamic equations (3) and
TABLE I
SPECIFICATION OF THE SYNRM UNDER TEST
(4) can be written in integral form as
t
λd = [vd (τ ) − id (τ )Rs ]dτ (7)
0
t
λq = [vq (τ ) − iq (τ )Rs ]dτ. (8)
0
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
THIKE AND PILLAY: AUTOMATED CURRENT CONTROL METHOD FOR FLUX-LINKAGE MEASUREMENT OF SynRMs 1467
TABLE II
PERCENTAGE ERROR IN CALCULATED INDUCTANCES WITH
VARIOUS SAMPLING TIMES
Fig. 2. Effect of dead time on the inverter output voltage showing output
voltage waveform for cases Io > 0 and Io < 0.
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
1468 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 2, MARCH/APRIL 2020
Fig. 7. Two control structures to ensure zero steady-state error. (a) PI con-
troller. (b) PDF controller.
Fig. 6. Schematics of the proposed control strategy to measure the d-axis flux
linkage and inductance (i∗q would be pulsed and i∗d would be constant for the
q-axis flux linkage and inductance measurement). Note: PDF stands for the Based on the mathematical model described by (3) and (4),
pseudoderivative feedback controller. the controlled systems are first-order systems. In order to ensure
zero steady-state error, the controllers must have an integrator.
(50 μs) is closer to the actual inductance compared to the induc- Fig. 7(a) and (b) shows the two controllers viz., PI and PDF
tances obtained from VPM with a lower sampling time (25 μs). controllers having only two controller parameters that assure
zero steady-state error for a first-order system. The closed-
IV. CURRENT CONTROL BASED METHOD loop transfer function for them is given by (10) and (11),
respectively. Standard form representations of these closed-loop
In order to overcome the aforementioned shortcomings, this transfer functions are shown in (12) and (13); where ωc is
article proposes a current control based method to measure the the closed-loop zero with the PI controller; ωnPI and ζPI are
flux-linkage and inductance characteristics of a SynRM. In the the natural frequency and the damping ratio of the closed-loop
proposed method, a pulsed current reference is applied in one system with PI controller, respectively; and ωnPDF and ζPDF are
axis while the current in the other axis is controlled to a constant. the natural frequency and the damping ratio of the closed-loop
Since both the currents are controlled, it is possible to measure system with the PDF controller, respectively. The expressions
the flux-linkage characteristics at any arbitrary operating point relating these standard form parameters to the controller and
in id iq plane. Additionally, due to the closed-loop control, the system parameters are given by (14) to (16)
accuracy in the measurement can be improved by modifying Kp
the response such that a sufficient number of data are available Io (s) Ki s + 1
= L (10)
during the transient. This is important in the implementation Iref (s) PI R+Kp
Ki s + Ki s + 1
2
using a real-time system with a limited sampling period.
Fig. 6 shows the schematic of the proposed technique to Io (s) 1
= L (11)
measure the d-axis inductances. In the proposed method, the Iref (s) PDF s 2 + R+Kp s + 1
Ki Ki
rotor is locked at a fixed position and a pulsed reference id is
Io (s) ωc s + 1
1
applied while iq is controlled to a fixed value. The three-phase
= 1 (12)
currents are measured and transformed to the dq frame. The Iref (s) PI(standard) 2 s2 + ω2ζnPI
PI
s+1
ωnPI
measured currents are subtracted from the references, and the
errors are passed through the controller to generate the required Io (s) 1
= 1 (13)
voltage signals. A three-phase VSI is used to apply the desired Iref (s) PDF(standard) s 2 + 2ζPDF s + 1
ω2
nPDF
ωnPDF
voltages in pulsewidth modulated form. In order to keep the fixed
current intact due to the pulsed current, the controller controlling Ki
ωc = (14)
the fixed current is made much faster compared to the controller Kp
controlling the pulsed current. Additionally, in order to meet the
Ki
accuracy requirement, the controller parameters for the pulsed ωnPI = ωnPDF = (15)
current controller are tuned to ensure an adequate number of L
samples during the transient period. R + Kp
ζPI = ζPDF = √ . (16)
2 Ki L
A. Current Controllers The closed-loop transfer function (10) using a PI controller
The proposed method requires two current controllers. The has a zero at ωc that depends on the controller parameters. The
major issue with the control problem is that the control design effect of this zero is the reduction in the rise time and increase
requires the parameters that are going to be measured. Thus, con- in the speed of response. If ωc is less than ωnPI , it causes
trol design is a challenging task in the proposed method. Another overshoot in the system that depends on the degree to which
requirement with the controllers is that the controlled parame- ωc is smaller, even though the closed-loop poles are critically
ters should have almost the same settling time irrespective of damped. Since the machine parameters vary due to saturation
the variation in the plant parameters (saturation effects). This and cross-coupling effects, the presence of a closed-loop zero
ensures the same accuracy in the measurement at all operating imposes an extra constraint to the controller design, and thus,
points. Moreover, the controller should be simple with the least it is undesirable. Additionally, the gain parameter Kp in the
possible number of parameters, yet ensuring the applicability of PI controller cannot be negative, otherwise, the initial response
an automatic tuning procedure. would go in the opposite direction until the voltage command
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
THIKE AND PILLAY: AUTOMATED CURRENT CONTROL METHOD FOR FLUX-LINKAGE MEASUREMENT OF SynRMs 1469
generated by integral gain becomes higher than the proportional TABLE III
EFFECT OF VARIATION IN INDUCTANCE ON ωn AND ζ
part. In contrast, the PDF controller has a parameter Kp that acts
similar to the feedback resistance [see Fig. 7(b)]. The response
with a negative value of Kp would be as if the resistance R is
reduced, which means a sluggish response. Therefore, due to
these reasons, the PDF controller is selected.
B. Controller Design
As discussed earlier, the controller should be able to increase
the speed of response for a slower system and reduce the speed
of response for a faster system. Based on the analysis presented
in Section III, the accuracy of measurement for a system with
a time constant of 100 ms at a sampling time of 25 μs is good
enough. Thus, in this article, one of the controller design criteria
is to have a response similar to such a system. The desired settling
time of the controlled system can be considered to be five times
this time constant. The design equations for the controller can
be obtained from (15) and (16); and they are given by (17) and
(18).
Ki = Lωn2 (17)
Kp = 2ζ Ki L − R = 2ζLωn − R. (18)
These equations show that the calculation of controller gains
require information on the machine resistance and inductance.
The system damping ratio, ζ and the natural frequency of oscilla- Fig. 8. Step response of a system with the same controller when its inductance
varies from nominal value of 100 to 5 mH and 200 mH. (a), (c), and (e) Controller
tion, ωn are defined explicitly based on the settling time. As seen outputs. (b), (d), and (f) Output currents.
from (15) and (16), these parameters change with the change in
system parameters. In addition, when ζ falls below one, there saturation and cross-coupling effects, only the inaccuracy in L
will be an overshoot, which means that the transient has a faster is analyzed. Consider that the actual inductance is ‘n’ times the
rise as well as oscillations. In order to avoid such oscillations, ζ estimated value. It may be due to estimation inaccuracy or due
should be remained greater than one even for extreme cases. to saturation effects. Relations (15) and (16) are used to see its
When ζ is greater than unity, the location of closed-loop poles effect in the closed loop ζ and ωn .
are given by (19). Table III presents new values of ζ and ωn with respect to the
p1,2 = −ζωn ± ωn ζ 2 − 1. (19) design specification for different values of “n.” It shows that
both the natural frequency and damping ratio varies when the
In order to define the values of ζ and ωn in terms of the actual inductance differs from the rough estimates. However,
required settling time (with no overshoot), the roots of the the ratio of ζ to ωn remained constant. In order to see the effect
characteristic equation in (13) are made far from each other. of variation in machine inductance on the closed-loop response,
When the value of ζ is two, the location of poles are at −3.732ωn the controller parameters are computed using the desired settling
and −0.268ωn . The first pole is much further from the origin time and nominal value of inductance. MATLAB Simulink is
than the second pole. Thus, the closed-loop system can effec- used to simulate the response of three different systems using the
tively be treated as a first order system with a time constant of same controller. For the test case, the nominal value of machine
1/(0.268ωn ). Considering the desired settling time to be five inductance is 100 mH and the settling time is chosen to be 1 s.
times of this time constant, the natural frequency ωn can be Fig. 8 compares the step responses with the same controller
calculated. for machine inductances of 100, 5, and 200 mH. It shows that
The remaining unknowns in (17) and (18) are L and R. In even though the machine inductance varies to a very high degree
order to obtain a rough estimate of these parameters, a pulsed from the rough estimates, the controller adjusts its output to
voltage of small magnitude that applies roughly 10%–20% of obtain almost the same response. Thus, a very rough estimate of
the rated current is applied to the machine, and (7) or (8) is inductance is enough for the controller design.
used to measure the machine resistance and inductance. These The controller design procedure can be summarized as fol-
estimates are rough, and can be highly inaccurate as discussed lows.
in Section III. The efficacy of these rough estimates on the con- 1) Choose the desired settling time based on the accuracy
troller performance is evaluated by analyzing the performance requirement.
of the controller when these parameters are varied. Since the 2) Compute the values of damping ratio and natural fre-
estimation in R is accurate enough and it is not affected by quency.
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
1470 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 2, MARCH/APRIL 2020
Fig. 10. Timing diagram of the pulsed reference current to calculate the
inductances online in a real-time processor.
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
THIKE AND PILLAY: AUTOMATED CURRENT CONTROL METHOD FOR FLUX-LINKAGE MEASUREMENT OF SynRMs 1471
Fig. 13. System response for an arbitrary operating point when a pulsed
reference id is applied at a fixed reference iq . (a) vd . (b) id . (c) vq . (d) iq .
phase currents are measured and fed back to the controller for
the closed-loop operation.
Fig. 13 presents the voltage and current waveforms for an
arbitrary operating point of id = 10 A and iq = 4 A. It shows
that the voltage vd applied by the controller during the transient
is higher compared to the steady-state vd . It is because the
machine time constant for this operating point is very large, and
an application of steady-state voltage from the beginning of the
pulse will slow down the response. The plot also shows that iq is
controlled to 4 A with a very small ripple. The ripple magnitude
depends on the switching frequency and dc bus voltage. In the
experiment, the switching frequency was 10 kHz, and the dc
Fig. 11. Flowchart showing the implementation of the proposed algorithm in bus voltage was reduced to 80 V compared to the nominal
a real-time system.
voltage of 350 V. The d-axis flux linkage at the operating point of
(id , iq ) = (10 A, 4 A) is obtained applying (20) to the measured
voltages and currents in the real-time processor. To obtain the
d-axis flux-linkage map for the entire dq plane, the reference
operating point is changed and the process is repeated. A similar
procedure is repeated to obtain the q-axis flux-linkage map.
Figs. 14 and 15 present the experimental results on the
variation of q-axis and d-axis flux linkages, respectively, with
id and iq . It shows that the q-axis flux linkage shows more
saturation effects than the d-axis flux linkage. Additionally, the
cross-coupling effects are more in the q-axis than the d-axis.
Comparing Fig. 14(a) and (c) while both present the q-axis flux
linkage curves for negative q-axis currents, the cross-coupling
Fig. 12. Experimental setup. (1) Test SynRM. (2) Dynamometer. (3) Real-time effects of id depend on both its magnitude and direction. For
controller. (4) Inverter. (5) Sensors. (6) Data acquisition system. the same magnitude of id , the change in the q-axis flux linkage
depends on its direction. A similar effect can be observed while
machine to obtain a rough estimate of the machine inductance comparing Fig. 14(b) and (d), Fig. 15(a) and (c), and Fig. 15(b)
and resistance. These values are used to calculate the controller and (d). In an ideal symmetric machine, the effect of id or iq on
parameters. As shown in Fig. 6, to measure the d-axis flux the cross-coupling flux linkage does not depend on its direction.
linkage and inductance, a pulsed reference current is applied This remains true for the SynRM with a symmetrical geometry.
in the d-axis while controlling the current in the q-axis to a The flux linkages obtained by averaging the flux linkages for
fixed value. PDF controllers are used to generate the references all the rotor positions must show symmetry. However, for any
vd∗ and vq∗ . The reference dq voltages are converted to the abc particular rotor position, it may not be true. The results presented
frame using a dq − abc transformation. An encoder is used to in Figs. 14 and 15 are for a particular rotor position. For some
get the position information for dq − abc transformation. The rotor positions, while magnetic flux in the machine is flowing
reference abc voltages are supplied to the pulsewidth modulation due to current in one axis, if current in another axis is applied,
generator to generate gating signals for the inverter. The three the resultant magnetic flux may depend on its direction. In order
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
1472 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 2, MARCH/APRIL 2020
Fig. 16. Comparison of Ld and Lq obtained using the proposed CCM with
the standard ac test.
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
THIKE AND PILLAY: AUTOMATED CURRENT CONTROL METHOD FOR FLUX-LINKAGE MEASUREMENT OF SynRMs 1473
VIII. CONCLUSION
Flux linkage and inductance measurement methods are evolv-
ing to meet the needs in machine and drive designs. To obtain
Fig. 17. Comparison of the d-axis flux linkage at two different magnitudes of the flux linkage and inductance map of a machine at standstill,
iq obtained using the proposed CCM with the CSM.
conventionally, a pulsed voltage is applied in one axis while
the current in another axis is controlled to remain constant.
This method requires a higher sampling rate for a machine
with a lower time constant and a longer pulse duration for a
machine with a higher time constant. This makes it unsuitable for
measurements using a real-time system. Moreover, the current
flowing in the d-axis induces a voltage in the q-axis due to the
inverter dead time and device drops. It puts a limitation to the
measurement range in the id iq plane.
This article proposed a novel current control based standstill
method to measure the flux linkage of a SynRM. In the proposed
method, instead of a voltage pulse as in the conventional test
method, a current pulse in closed loop is applied in one axis while
Fig. 18. Comparison of the q-axis flux linkage at two different magnitudes of controlling the current in another axis. Due to the use of current
id obtained using the proposed CCM with the CSM.
control, flux linkage at any operating point in dq plane can be
measured. In addition, the transient response of the machine can
also show that both the inductances Ld and Lq decreases with be modified such that adequate number of measurement samples
an increase in id and iq , respectively. is available during the transient, making it suitable to use in a
The dq axes flux linkages of the test motor including cross- real-time system for an online measurement of flux linkages.
saturation effects are also measured using the CSM. Details on This article also presented in detail the analysis and design of the
the underlying principle behind the CSM are described in [14], controllers required in the proposed method. Finally, this article
and this article followed the CSM presented in the same paper. presented an algorithm to implement the proposed method using
Figs. 17 and 18 present the comparison of dq flux linkages at two a real-time system. The proposed method is verified using an
different magnitudes of iq and id obtained using the proposed experiment performed on a test machine. In order to assess
CCM with the CSM. In Fig. 17, the d-axis flux linkage for the the efficacy of the proposed method, the results obtained from
same magnitude of id has reduced when iq is applied to the the proposed method are compared with the results obtained
machine. This reduction is due to saturation of the common from the standard ac standstill test and CSM. The results show
magnetic path by iq , called cross-magnetizing effect. A similar that the proposed method is useful to obtain the flux-linkage
effect is seen in Fig. 18 in the q-axis flux linkage when id is characteristics of a SynRM using a real-time system.
applied. There is a nominal difference in the magnitude of the
flux linkage obtained from CSM with the one obtained from REFERENCES
the proposed current pulse method. However, the difference is [1] X. Ma, G. J. Li, Z. Q. Zhu, G. W. Jewell, and J. Green, “Investigation
less than 2%. It should be noted that the CSM result is an av- on synchronous reluctance machines with different rotor topologies and
winding configurations,” IET Elect. Power Appl., vol. 12, no. 1, pp. 45–53,
erage flux linkage around rotor positions, whereas the proposed 2018.
method gives position-dependent flux linkages. Based on the [2] A. Vagati, M. Pastorelli, G. Francheschini, and S. C. Petrache, “Design
experiments performed at different rotor positions and current of low-torque-ripple synchronous reluctance motors,” IEEE Trans. Ind.
Appl., vol. 34, no. 4, pp. 758–765, Jul. 1998.
magnitudes, the variation in inductances with rotor position is [3] T. Matsuo and T. A. Lipo, “Rotor design optimization of synchronous
lower than 2% for Ld and 6% for Lq . For the comparison in reluctance machine,” IEEE Trans. Energy Convers., vol. 9, no. 2,
Figs. 17 and 18, the measurements at the rotor position with pp. 359–365, Jun. 1994.
[4] M. J. Kamper and A. F. Volsdhenk, “Effect of rotor dimensions and
rotor the d-axis aligned with stator A-phase axis is used. cross magnetisation on Ld and Lq inductances of reluctance synchronous
Thus, the proposed method is useful in measurements of the machine with cageless flux barrier rotor,” IEE Proc.—Elect. Power Appl.,
flux linkage and inductance maps using a real-time system. vol. 141, no. 4, pp. 213–220, Jul. 1994.
[5] S. Yamamoto, H. Hirahara, J. B. Adawey, T. Ara, and K. Matsuse,
This method does not require post processing of a huge dataset. “Maximum efficiency drives of synchronous reluctance motors by a novel
Compared to the pulsed voltage reference method, the proposed loss minimization controller with inductance estimator,” IEEE Trans. Ind.
method improves the accuracy for machines ranging from very Appl., vol. 49, no. 6, pp. 2543–2551, Nov. 2013.
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.
1474 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 56, NO. 2, MARCH/APRIL 2020
[6] M. Ferrari, N. Bianchi, and E. Fornasiero, “Analysis of rotor saturation in [26] N. R. Tavana and V. Dinavahi, “Real-time FPGA-based analytical space
synchronous reluctance and PM-assisted reluctance motors,” IEEE Trans. harmonic model of permanent magnet machines for hardware-in-the-
Ind. Appl., vol. 51, no. 1, pp. 169–177, Jan. 2015. loop simulation,” IEEE Trans. Magn., vol. 51, no. 8, Aug. 2015,
[7] T. Lubin, H. Razik, and A. Rezzoug, “Magnetic saturation effects on Art. no. 8106609.
the control of a synchronous reluctance machine,” IEEE Trans. Energy [27] C. Dufour, S. Cense, T. Yamada, R. Imamura, and J. Bélanger,
Convers., vol. 17, no. 3, pp. 356–362, Sep. 2002. “FPGA permanent magnet synchronous motor floating-point models
[8] A. Vagati, M. Pastorelli, F. Scapino, and G. Franceschini, “Impact of cross with variable-DQ and spatial harmonic finite-element analysis solvers,”
saturation in synchronous reluctance motors of the transverse-laminated in Proc. 15th Int. Power Electron. Motion Control Conf., Sep. 2012,
type,” IEEE Trans. Ind. Appl., vol. 36, no. 4, pp. 1039–1046, Jul. 2000. pp. LS6b.2-1–LS6b.2-10.
[9] R. Thike and P. Pillay, “Experimental investigation of MTPA trajectory [28] K. S. Amitkumar, R. S. Kaarthik, and P. Pillay, “A versatile power-
of synchronous reluctance machine,” in Proc. IEEE Int. Conf. Power hardware-in-the-loop-based emulator for rapid testing of transportation
Electron., Drives, Energy Syst., Chennai, India, Dec. 2018, pp. 1–6. electric drives,” IEEE Trans. Transp. Electrific., vol. 4, no. 4, pp. 901–911,
[10] C. M. Hackl, M. J. Kamper, J. Kullick, and J. Mitchell, “Current control of Dec. 2018.
reluctance synchronous machines with online adjustment of the controller [29] K. S. Amitkumar, R. Thike, and P. Pillay, “Linear amplifier-based power-
parameters,” in Proc. IEEE 25th Int. Symp. Ind. Electron., Jun. 2016, hardware-in-the-loop emulation of a variable flux machine,” IEEE Trans.
pp. 153–160. Ind. Appl., vol. 55, no. 5, pp. 4624–4632, Sep. 2019.
[11] M. N. Ibrahim, P. Sergeant, and E. M. Rashad, “Relevance of including sat- [30] J. Lee, Y. Kwon, and S. Sul, “Experimental identification of IPMSM
uration and position dependence in the inductances for accurate dynamic flux-linkage considering spatial harmonics for high-accuracy simulation of
modeling and control of SynRMs,” IEEE Trans. Ind. Appl., vol. 53, no. 1, IPMSM drives,” in Proc. IEEE Energy Convers. Congr. Expo., Sep. 2018,
pp. 151–160, Jan. 2017. pp. 5804–5809.
[12] K. S. Amitkumar, R. Thike, and P. Pillay, “Power-hardware-in-the-loop [31] S.-H. Hwang, J.-M. Kim, H. V. Khang, and J.-W. Ahn, “Parameter iden-
based emulation of a variable flux machine,” in Proc. IEEE Energy tification of a synchronous reluctance motor by using a synchronous PI
Convers. Congr. Expo., Sep. 2018, pp. 6454–6460. current regulator at a standstill,” J. Power Electron., vol. 10, pp. 491–497,
[13] R. Nuscheler, “Determination of the inductances and the rotor flux-linkage Sep. 2010.
of a permanent-field synchronous machine with reluctance effects,” in [32] R. Thike and P. Pillay, “Automatic inductance measurements of syn-
Proc. 4th Int. Power Electron. Motion Control Conf., Aug. 2004, vol. 2, chronous reluctance machines including cross-saturation using real-time
pp. 620–625. systems,” in Proc. IEEE Energy Convers. Congr. Expo., Sep. 2018,
[14] S. Wiedemann, S. Hall, R. M. Kennel, and M. Alaküla, “Dynamic testing pp. 6121–6127.
characterization of a synchronous reluctance machine,” IEEE Trans. Ind. [33] IEEE Draft Trial-Use Guide for Testing Permanent Magnet Machines,
Appl., vol. 54, no. 2, pp. 1370–1378, Mar. 2018. IEEE P1812/D2, Feb. 22, 2013.
[15] R. Thike and P. Pillay, “Vector controlled drive to measure inductances of
variable flux machine,” in Proc. IEEE Int. Conf. Power Electron., Drives,
Energy Syst., Dec. 2016, pp. 1–6.
[16] S. Yamamoto, T. Kano, Y. Yamaguchi, and T. Ara, “A method to determine
direct- and quadrature-axis inductances of permanent magnet synchronous
motors,” Elect. Eng. Jpn., vol. 171, no. 3, pp. 41–50, 2010. [Online].
Available: https://onlinelibrary.wiley.com/doi/abs/10.1002/eej.20969 Rajendra Thike (Student Member, IEEE) received
[17] Ž. Plantić and G. Štumberger, “Determining parameters of a three the bachelor’s degree in electrical engineering from
phase permanent magnet synchronous machine using controlled single- Tribhuvan University, Kathmandu, Nepal, in 2010,
phase voltage source,” Przeglad Elektrotechniczny (Elect. Rev.), vol. 87, and the master’s degree in electrical engineering
pp. 137–140, 2011. from Concordia University, Montréal, QC, Canada,
[18] J. B. Im, W. Kim, K. Kim, C. S. Jin, J. H. Choi, and J. Lee, “Induc- in 2016. He is currently working toward the Ph.D.
tance calculation method of synchronous reluctance motor including iron degree in electrical engineering with the Department
loss and cross magnetic saturation,” IEEE Trans. Magn., vol. 45, no. 6, of Electrical and Computer Engineering, Concordia
pp. 2803–2806, Jun. 2009. University, Montréal, QC, Canada.
[19] S. Yamamoto, T. Ara, and K. Matsuse, “A method to calculate transient He was an Electrical Engineer with the Ministry of
characteristics of synchronous reluctance motors considering iron loss Energy, Government of Nepal.
and cross-magnetic saturation,” IEEE Trans. Ind. Appl., vol. 43, no. 1,
pp. 47–56, Jan. 2007.
[20] A. Kilthau and J. M. Pacas, “Parameter-measurement and control of the
synchronous reluctance machine including cross saturation,” in Proc. Conf.
Rec. IEEE Ind. Appl. Conf., 36th IAS Annu. Meeting, Sep. 2001, vol. 4,
pp. 2302–2309.
[21] B. Stumberger, G. Stumberger, D. Dolinar, A. Hamler, and M. Trlep,
“Evaluation of saturation and cross-magnetization effects in interior
permanent-magnet synchronous motor,” IEEE Trans. Ind. Appl., vol. 39, Pragasen Pillay (Fellow, IEEE) received the bach-
no. 5, pp. 1264–1271, Sep. 2003. elor’s and master’s degrees in electrical engineer-
[22] R. Thike and P. Pillay, “Characterization of a variable flux machine ing from the University of KwaZulu-Natal, Durban,
for transportation using a vector-controlled drive,” IEEE Trans. Transp. South Africa, in 1981 and 1983, respectively, and the
Electrific., vol. 4, no. 2, pp. 494–505, Jun. 2018. Ph.D. degree in electrical engineering from Virginia
[23] N. Bedetti, S. Calligaro, and R. Petrella, “Stand-still self-identification Polytechnic Institute and State University, Blacks-
of flux characteristics for synchronous reluctance machines using novel burg, VA, USA, in 1987.
saturation approximating function and multiple linear regression,” IEEE He is currently a Professor with the Department
Trans. Ind. Appl., vol. 52, no. 4, pp. 3083–3092, Jul. 2016. of Electrical and Computer Engineering, Concor-
[24] M. Hinkkanen, P. Pescetto, E. Mölsä, S. E. Saarakkala, G. Pellegrino, dia University, Montréal, QC, Canada, where he is
and R. Bojoi, “Sensorless self-commissioning of synchronous reluctance with the Natural Sciences and Engineering Research
motors at standstill without rotor locking,” IEEE Trans. Ind. Appl., vol. 53, Council of Canada, Ottawa, ON, Canada/Hydro-Québec Senior Industrial Re-
no. 3, pp. 2120–2129, May 2017. search Chair, Montréal, QC, Canada.
[25] N. R. Tavana and V. Dinavahi, “A general framework for FPGA-based Dr. Pillay was the recipient of the 2015 IEEE IAS Outstanding Achievement
real-time emulation of electrical machines for HIL applications,” IEEE Award. In 2017, he was appointed as a Honorary Professor with the University
Trans. Ind. Electron., vol. 62, no. 4, pp. 2041–2053, Apr. 2015. of Cape Town, Rondebosch, South Africa.
Authorized licensed use limited to: Tokyo Institute of Technology. Downloaded on November 21,2023 at 06:03:55 UTC from IEEE Xplore. Restrictions apply.