High Performance PM Synchronous Motor Drive
for an Electrical Scooter
Bianchi N., Bolognani S., Zigliotto M.
Department of Electrical Engineering, University of Padova Via Gradenigo, 6a, 3513 I Padova, Italy
fax: +39 .049 . 827 7599, e-mail bolognan@dei.unipd.it
Abstract - In this paper, the design of a full electrical
system for an electrical scooter is presented. The paper
explains the design choices that have been done to
satisfy the typical main requirements for an electrical
vehicle. According to them, an interior permanent
magnet synchronous motor has been incorporated in a
drive able to control the motor both in the constant
torque (constant flux) and in the constant volt-ampere
(flux-weakening) regions. Experimental results carried
out on a system prototype are also included in the final
paper.
an anisotropic rotor, with a fairly reduced PM flux. A
commercial 24-slot induction motor stator is used, with a
single-layer winding. Fig.1 shows two pole-pitches of the
motor.
x barrier
I. INTRODUCTION
Increasing awareness of air quality and interest in
innovative vehicles stimulate research activity to improve
the propulsion systems by reducing the vehicle .emissions.
In recent years, considerable developments have been
obtained in the area of electric and hybrid vehicles, whose
use solves the environmental problems caused by the use of
internal combustion engine vehicles. Moreover, the torque
generated by the electric motor can be appropriately
controlled so that the vehicle stability and safety result
greatly improved. In this paper, the design of a full
electrical system for an electrical scooter is presented. The
paper explains the design choices that have been done in
order to satisfy the typical main requirements for an
electrical vehicle, namely:
wide flux-weakening (FW) region, to minimise the drive
power rating in spite of a wide speed range;
high-efficiency to increase the vehicle autonomy;
high torque-to-ampere and torque-to-volume ratios to
minimise the inverter cost and the drive overall
dimensions;
simple manufacture for a cost-effective end product.
Several considerations are given explaining the design
choices for both motor and control. Experimental tests
validate the theoretical results.
11. SYNCHRONOUS MOTOR DESIGN
The aforementioned goals can be achieved by the use of a
synchronous Interior Permanent Magnet (IPM) motor drive,
even in case of FW, as widely reported in the specialised
literature [ 1-51. The most suitable motor structure considers
Fig.1- Schematic ofthe interior PM motor
A. Preliminary design
The design of the IPM motor for FW operations is normally
a two-step procedure. Firstly, the design requires to
determine a set of motor parameters that satisfies the
required performance in the whole operating range. The
parameters used during this first rough design step are
usually normalised [6-81, as reported in Appendix. In the
following, normalised parameters will be written in italic
with small letters. Secondly, the actual motor sizing is
developed from the normalised parameters issued by the
first step. The procedure has been already described in [9-
101, and it will be here omitted. On the contrary, some
considerations on the first design step are reported in the
following. It can be fbrther subdivided into three phases,
listed hereafter:
The saliency ratio €,=ldld is first selected, in accordance
with typical values of the adopted rotor structure (Fig. 1).
In the present work, the range is between 3 and 5, and an
initial guess value 5=4 has been chosen. As clarified
below, further coefficients will be futed to consider the
saturation effect.
Starting ffom the specifications on FW speed & and
torque z j j , the suitable PM flux linkage A,,, is chosen.
From 5 and ;Im the corresponding normalised q and d
axis inductance lq, Id and the rated stator current iiy are
computed.
The design parameters have to comply with torque and
voltage specifications, as well as with maximum torque-to-
current ratio relationship [lo]. Even if there is a certain
0-7803-6401-5/00/$10.00 0 2000 IEEE
1901
degree of freedom in the choice, it is important to recall
that different combinations of iq, iN and Am do not yield to
the same FW performance. Furthermore, the effects of
motor non-linearity, due to iron saturation, has to be taken
into account [ll-121.
For the actual application, a constant power for a speed
range equal to three, i.e. power pfiu"l at 9 = 3 (thus
r/,.=l>.33 at %=3) and, of course, f b = l at cwl were
required.
A simple piecewise linear model, that deals with the q-axis
saturation, has been introduced and verified by the authors
in [13]. It considers a constant PM flux linkage and a
constant d-axis inductance, together with two values of the
q-axis inductance: the unsaturated lqu and the saturated
(differential) fqs . In particular, it has been demonstrated that
a saturated motor has to be designed with higher PM flux
linkage and lower inductance than the corresponding
unsarurated motor. The model has been used in [14] to
obtain the normalised motor parameters.
F i g 2 reports the FW torque rfi and power ( z j j 9 ) at FW
speed ~ q i ~ 3 as functions of the PM flux linkage and ratios
<=iqJld and (s=iqJfqs. It is worth to highlight that cs is
related to the magnetic characteristics of the q-axis path and
that (,=1 refers to an unsaturated motor. Typical values for
the selected configuration range from 5,=3 to 7 and here an
initial guess 5,=5 has been chosen, as reported in Fig.2. The
curves shown also depend on the ratio o,=iq.diq, where iqs is
roughly the current at which saturation begins while iq is the
rated base current. It is thus evident that oq indicates the
saturation level: a unity value refers to an unsaturated
motor, and hence a poor exploitation of the magnetic paths,
while lower values indicate an higher saturation level [14],
that can yield higher iron losses. Winding thermal limits
also comes into play if too low oq are selected. Typical
trade-off values are from 1 down to 0.4.
I;. FW power and FW torque ?li"
1 s
1.35
1 2
1 os
0 9
0.75
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
A m
Fig.2 - FW performances, at different saturation levels
The experimental prototype developed for this work has
been indicated with a black mark on Fig2 and Fig.3.
Obviously, it is the consequence not only of the given
specifications, but also of the intermediate design choices
of & and or. Fig.2 returns a value of the PM flux linkage
that can be entered into other two curves obtained from the
same simplified model [I41 and reported in Fig.3 (a) and
(b).
unsaturated q-axis inductance
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
A m
rated current
0 0.1 0.2 0.3 0.4 0.5 0.6 07 0.8 0.9 1
A m
Fig3 - (a) Normalised q-axis inductance and (b) stator current
These characteristics are then exploited to get the values of
I,, and IN. Fig.4 resumes the design steps outlined above.
Fig.4 - Choice of the normalised IPM motor parameters
111. ANALYSIS AND EXPERIMENTAL MEASUREMENTS ON
THE MOTOR PROTOTYPE
This Section presents some results obtained from Finite
Element (FE) analysis on the motor prototype. They are
then validated by means of experimental measurements.
80
70
60
50
40 current
30
20
10
q-axis
(A)
-80 -70 -60 -50 -40 -30 -20 -10 0
d-axis current (A)
Fig.5 - Simulated and experimental results: constant torque loci
1902
Wlth reference to the designed IPM motor, Fig.5 shows the
constant-torque loci in the plane (Id-L,), solid l i e .
Three current-limit (peak value) circles are drawn in dashed
line Black dots in the figure indicate the measured values
of motor torque, at the corresponding current operating
point. F ig5 highlights that the measured results are in
agreement with the simulated ones.
I t I S well known that a higher efficiency can be achieved in
each drive working condition by feeding the motor
according to the maximum torque-to-current ratio strategy.
The effective match between FE analysis and experiments
enables the use of the former to track the maximum torque-
to-current ratio trajectory shown with bold line in Fig.5. AS
a result, Fig.6 reports the d-axis and q-axis currents as
functions of the stator current magnitude is=1iJ,
particularised for the actual parameters of the motor
prototype.
Y Y
0 IO 20 30 40 50 60 70 80 90
current magnitude (A)
I-ig.6 - The d- and q-axis current components for maximum
torque-to-current ratio (actual base point of operations, not
normalised)
Iv. DRIVE STRUCTURE DESIGN
The application of the IPM motor drive to an electrical
scooter allows enlarged speed range in the constant power
conditions, with an inverter size lower than in a
conventional flux-oriented induction motor drive [ 1,2]. AS
a counterpart, the presence of both reluctance and PM
torque requires involved control strategies, to guarantee
effective drive operations at all operating conditions. The
block schematic of the complete drive structure is sketched
in Fig.7.
The throttle knob position (tkp*) is here considered as a
speed reference, and it is compared with the actual
mechanical speed a,, as shown in Fig.7. A light pure
proportional control P then delivers the stator current
reference space vector is*.
The speed loop aims to emulate the behaviour of a fuel
engine scooter, in which there is a steady state speed error
proportional to the load, i.e. the longitudinal slope of the
road
SVM inverter
Id
Fig.7 - IPM motor drive block schematic
The main issues for an efficient control are a proper current
settings delivery and the avoidance of any inverter voltage
saturation. Both aspects will be examined carefully in the
next sections.
A. The current reference generator
The amplitude of stator current i, is directly linked to the
torque production. Nevertheless, there are infinite couples
id, i, that give the same vector is, each producing a different
torque value. In the constant torque operating range various
profiles in the id, i, plane can be selected, depending on the
optimisation objective of the specific application [ 151. In
this work the maximum torque per ampere strategy has
been adopted, due to its simplicity and operational
significance [ 161.
For a given id, i, pair and under the hypothesis of system
linearity the electromagnetic torque is given by
T =-pi,[A,-(L,-Ld)id] 3 .
2
If the direct current reference id* is expressed as function of
i,'and the current reference amplitude is*
it is possible to determine the value of the reference i,* that
maximises the torque, by imposing
that gives in turn
1903
in which
resistive drops have been neglected. It is then
is the electrical speed (pw,) and the stator
i* (1 = sign(r) (4)
(7 )
where ida=Am/(Ld'Lq).
It is soon evident that, in presence of deep saturation of the
q-path, expressions (2) and (4) are only rough and ready
approximations of the correct d-q current references. Their
use implies less efficient current settings for a given torque
request, with increased energy consumption.
In this work, FE method has produced accurate estimations
of d-q axes current references as function of stator current
amplitude (Fig.6). The analysis was based on the search of
the couple id, iq that maximises the torque, now calculated
by
i; =
me
It iS worth to note that the theoretical Value (7) is function
of either current and voltage vector amplitude and of the
motor speed as well. The use of (7) as current reference
generator requires the knowledge of the electrical speed of
in which the FW has to be triggered, given by
where hd'hd(id) and hq=hq(iq). The d-q axis cross-
coupling, i.e. the effect of each current on the other axis, is
here neglected, since it yields a few percents error in the
torque computed by (5). The torque value ( 5 ) compared
with (1) is more accurate since the former holds also in
presence of magnetic non-linearity.
For each quantified is* value, the id, iq references of Fig.6
have been stored in a look-up table, that can be used
directly as optimal current reference generator in the
constant torque region (Fig.7).
The extended torque-speed operation of the IPM motor can
be reached by increasing negative values of the direct
current component, that in turn produces a d-axis flux
linkage opposite to the PM flux linkage (FW control) [7].
The generation of the additional FW direct current
reference id; is not a trivial task, since it results from a
trade-off between the best exploitation of the constant-
torque range and a prompt voltage fimitation prevention.
That is, an early or too strong FW action would reduce for
no reason the available maximum torque, while with a
delayed or too weak action the voltage limits would be
overcome, with subsequent cross-coupling effects on the
current regulation loops and possible instability.
A common way to derive the adequate current references
for FW is based on the calculation of the intersection
between the actual current circle and the voltage limit
ellipse in the id-& plane [17].
For a given current is* and a rated amplitude Ulim of
voltage vector the id*, i,* couple can be derived from (2)
and the steady state voltage balance equation
Since current measurements are available, it is possible, in
principle, to use the FE data to obtain the exact values of
the inductances Ld and L,, also considering the saturation
effects. These values can be used to decide by (8) whether
the FW strategy (7) has to be started or not. It is clearly a
cumbersome procedure, that takes up a considerable
amount of processing time. It is also noteworthy that during
transient operation below wfthe voltage reference vector ua
'+jup* could still exceed the available inverter voltage, and
the actual currents could not follow the references, with
performance degradation [ 151.
All the above indicates that automatic FW triggering and
current setting modification would be a great improvement,
especially in battery-energised applications in which every
waste must be carehlly avoided. An effective solution is
the realisation of an outer voltage loop, (Fig.7), which
delivers a FW component id< that is added to the actual id*
when the SVM (space vector modulated) inverter voltage
limit is approached. The quadrature current reference i,' is
now limited by
(9)
This solution was first proposed by 1171. Here it is
presented an improved version with some innovative details
that sensibly improve the system performance under
particularly stressing operational conditions.
1904
B. The voltage control loop
First of all, an accurate linearisation around the voltage
limit working point has confirmed the experimental
evidence that the proportional part greatly reduce the
stability of the voltage loop, and therefore a pure integral
action is to be preferred. The design now focuses on the
choice of KIu and Kv, that fixes the reference voltage limit
of the loop (Fig.7).
The maximum pure sinusoidal voltage vector that can be
produced for a given dc link is O.577*Udc. Of course, K,
must be less than 0.577, and in this work a value K,=0.54
has been used. It must be mentioned that exceedingly low
values trigger the FW in advance, as shown in case (b) of
Fig.8.
6 ' t
5 .
4 .
3 -
2 -
1 -
OL I
0 100 200 300 400 500 600
w , (radls)
Fig.8 - IPM drive torque-speed behaviour.
(a) without FW (b) Kv=0.5 (c) Kv=0.54
Fig.8 reports the actual drive torque-speed behaviour. As
expected, in absence of any FW action the voltage limit is
soon reached and the delivered torque falls almost
immediately as the base speed is exceeded, while an
accurate choice of K, yields the full exploitation of the
constant torque range as reported in case (c).
It has also been found that by using a pure integral action,
the choice of the gain constant Ki, is not critical. In
complete saturation condition, the actual voltage vector
traces the hexagonal boundary. The maximum (negative)
input error is given by the difference between the actual
voltage amplitude and the fixed limit Kv*Udc (Fig.7). As
the FW action starts by producing an increasing idf
Component, the error decreases to zero.
As a simple tule of thumb, it is reasonable to choose K;, so
thal in presence of the maximum input error the rated FW
component i&;N is delivered within a fixed time, let's say 30
t imes the id current loop rise time tri. Since the amplitude of
ihe fundamental voltage in case of hexagonal saturation is
0.6056 u d c [IS], K;, can be calculated as
K. = Idf ,N
30t,; (0.6056- K,)Udc IU
Several simulation results have confirmed that if K,, is
chosen sensibly smaller than (10) the voltage loop is no
more able to follow the transient voltage produced by the
current control and inverter saturation occurs. Conversely,
if the voltage loop is too fast, instability due to system
delays have been experienced.
Another crucial design item is the maximum negative FW
current whose value has to be chosen according to
the actual working point by
where IN is the rated stator current and idt is the direct
current that the current reference generator would give in
absence of voltage limitation. Actually, if a fixed limitation
would be chosen instead, instability may arise at load
detaching during deep FW operations. Fig9 illustrates the
principle.
-\
voltage current
limit
cimstant
limit
'df,N mdfN\;/
idt 1
Fig.9 - Voltage loop output limitation design
Let's consider point A the base working point, and assume
that a constant maximum FW current idf,mln=idf,N is
selected. If the IPM is accelerated up to two times the base
speed, let point B be the actual intersection between the
voltage limit and the stator current required by the speed
loop, At steady state, the inertia torque component drops
off sharply and the stator current request also reduces
(small circle in Fig.9).
Due to the given idf,min limitation, point C becomes the
reference working point, and since it lays outside the
voltage limit ellipse, the current control could become
unstable. Conversely, if (10) is used the integral loop reacts
and the final stable point D can be reached. Fig.10 shows
the different FW components generated with fixed output
limitation (case 1) and using (1 0) (case 2, dashed line).
1905
-do1 -60
b 1 2 3 4 5 k-*O0 -80
f (SI
Fig.10 - FW action at load torque detaching
Fig.11 and Fig.12 show the effects on different FW actions
on current control. In particular, with fixed limitation (case
I ) both control loop loose current control, and oscillations
arise around the final steady state working point. Perfect
current control is maintained using the proposed method
(case 2).
-80
0 1 2 3 4 5 6
Fig. I I - id current control at load torque detaching
J
t (SI
30.
20
10.
0 .
-'Ob 1 2 3 4 5 L
t (SI
Fig. 12 - i 4 current control at load torque detaching
The calculation of the actual stator voltage is based on
voltage references (Fig.7). To avoid the imprecision
essentially due to dead times and voltage drops on
switching devices, an appropriate dead times compensation
(DT cmp) algorithm [ 191 has been implemented.
Finally, the voltage saturation can also be sensed simply by
monitoring the period To of application of the zero-voltage
vector, as delivered by the SVM algorithm.
C. Management of current regulators saturation levels
In every modern AC drive the current control plays a
crucial rule and can be fully considered a distinguished
feature of the drive itself. Actually, a proper current control
can be achieved in the synchronous d-q coordinates system
by conventional PI contr
本文档为【High Performance PM Synchronous Motor Drive】,请使用软件OFFICE或WPS软件打开。作品中的文字与图均可以修改和编辑,
图片更改请在作品中右键图片并更换,文字修改请直接点击文字进行修改,也可以新增和删除文档中的内容。
该文档来自用户分享,如有侵权行为请发邮件ishare@vip.sina.com联系网站客服,我们会及时删除。
[版权声明] 本站所有资料为用户分享产生,若发现您的权利被侵害,请联系客服邮件isharekefu@iask.cn,我们尽快处理。
本作品所展示的图片、画像、字体、音乐的版权可能需版权方额外授权,请谨慎使用。
网站提供的党政主题相关内容(国旗、国徽、党徽..)目的在于配合国家政策宣传,仅限个人学习分享使用,禁止用于任何广告和商用目的。