IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO.

1, JANUARY 2011 3

Letters
Digital Implementation of an Adaptive Speed Regulator for a PMSM
Han Ho Choi, Nga Thi-Thuy Vu, and Jin-Woo Jung

Abstract—We design an adaptive speed regulator for a Finally, simulation and experimental results are shown to verify
permanent-magnet synchronous motor (PMSM). The proposed that the proposed method can precisely control the speed of a
adaptive regulator does not require any information on the PMSM PMSM under model parameter and load-torque variations.
parameter and load-torque values, thus, it is insensitive to model
parameter and load-torque variations. We implement the proposed
adaptive-speed-regulator system by using a TMS320F28335 float- II. MODEL DESCRIPTION
ing point DSP. We give simulation and experimental results to ver-
ify that our method can be successfully used to control a PMSM By taking the rotor coordinates of the motor as reference
under model parameter and load-torque variations. coordinates, a surface-mounted PMSM can be represented by
Index Terms—Adaptive-control system, disturbance, the following nonlinear equation:
permanent-magnet synchronous motor (PMSM), uncertainty. ⎧
⎨ ω̇ = k1 iq s − k2 ω − k3 TL
i̇ = −k4 iq s − k5 ω − ωids + k6 Vq s + dq (1)
I. INTRODUCTION ⎩ qs
i̇ds = −k4 ids + ωiq s + k6 Vds + dd
PERMANENT-MAGNET synchronous motor (PMSM)
A has been widely employed in servo applications such as
chip-mount machines, robotics, and hard disk drives because it
where TL represents the load torque, ω is the electrical rotor
angular speed, iq s is the q-axis current, Vq s is the q-axis voltage,
features low noise, low inertia, high efficiency, and low main- ids is the d-axis current, Vds is the d-axis voltage, dq and dd
tenance cost. However, a PMSM cannot be easily controlled represent disturbance inputs, and ki > 0, i = 1, . . . , 6 are the
because of the uncertainties such as parameter variations and parameter values depending on the number of poles, the stator
load-torque variations. Therefore, the linear-control methods resistance, the stator inductance, the rotor inertia, the viscous
such as PID control cannot guarantee high performance. To friction coefficient, and the magnetic flux. The load-torque dis-
solve this problem, many researchers have proposed various de- turbance term TL and uncertainties on the parameters ki can
sign methods, e.g., adaptive control [1]–[4], nonlinear feedback severely deteriorate the control performance if they are not ap-
linearization control [5], and fuzzy control [6], [7]. Recently, propriately accounted for.
several authors [8]–[10] have proposed disturbance-observer- We will use the following assumptions:
based PMSM control methods that can effectively suppress A1: ω, iq s , and ids are available.
load-torque variations. However, most of the previous PMSM A2: The desired speed ωd is constant and ω̇d = ω̈d = 0.
control design methods cannot guarantee stability and conver- A3: TL is unknown, ṪL can be set as ṪL = 0.
gence of speed error responses under inexact information on A4: ki are not known accurately, k̇i can be set as k̇i = 0.
the PMSM parameters such as the stator resistance, the sta- A5: dq and dd are unknown, d˙q and d˙d can be set as
tor inductance, the rotor inertia, the viscous friction coefficient, d˙q = d˙d = 0.
the magnetic flux, etc. Considering these facts, we propose an We will denote the electrical rotor angular acceleration ω̇ by
adaptive-control law design method for a PMSM. We first design β = k1 iq s − k2 ω − k3 TL = ω̇. Then, by introducing the speed
an adaptive speed regulator which does not require any informa- error ωe = ω − ωd and by denoting Vq f and Vdf as
tion on the PMSM parameter and load-torque values. We also ⎧ 1
prove that the speed error of the closed-loop system converges ⎪
⎪ Vq f = (k4 iq s + k5 ω + ωids − dq )
⎪
⎪ k
to zero and the parameter adaptation error signal is bounded. ⎨ 6
1
+ [k2 − γq ]β (2)
⎪
⎪ k1 k6
⎪
⎪
Manuscript received March 17, 2010; revised May 25, 2010; accepted June ⎩ V = 1 (k i − ωi − d )
23, 2010. Date of current version December 27, 2010. This work was supported df 4 ds qs d
k6
by Basic Science Research Program through the National Research Foundation
(NRF) of Korea funded by the Ministry of Education, Science, and Technology we can obtain the following error dynamics
( 2010-0015428) and by the Dongguk Research Fund. Recommended for pub-
lication by Associate Editor R. Burgos. ⎧
The authors are with the Division of Electronics and Electrical Engineer- ⎨ ω̇e = β
ing, Dongguk University - Seoul, Seoul 100-715, Korea (e-mail: jinwjung@ β̇ = −γq β + k1 k6 (Vq s − Vq f ) (3)
dongguk.edu). ⎩
Color versions of one or more of the figures in this paper are available online i̇ds = k6 (Vds − Vdf )
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TPEL.2010.2055890 where γq > 0 is a constant design parameter.

0885-8993/$26.00 © 2010 IEEE

4 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 1, JANUARY 2011

After all, our design problem can be formulated as designing therefore, (8) can be reduced to
an adaptive control law [Vq s , Vds ]T for the aforementioned error
dynamics (3). 2V̇ ≤ −k1 k6 δq σq2 − δd k6 i2ds ≤ 0 (12)
Before proceeding further, we give a background result which where ki > 0, δq > 0, and δd > 0 are used. The inequality (12)
will be used to derive our main results. implies that
Lemma 1: There exist constant parameter vectors ξq∗ = ∞ ∞ ∞
[ξq 1 , ξq∗2 , ξq∗3 , ξq∗4 ]T and ξd∗ = [ξd1
∗ ∗ ∗
, ξd2 ∗ T
, ξd3 ] such that 2 V̇ (τ )dτ ≤ −k1 k6 δq σq dτ − δd k6
2
i2ds dτ.
0 0 0

4 
3
(13)
hTq ξq∗ = hq i ξq∗i = Vq f , hTd ξd∗ = ∗
hdi ξdi = Vdf (4) Multiplying (13) by −1 and integrating the left-hand side of
i=1 i=1
(13) give
where hq = [hq 1 , hq 2 , hq 3 , hq 4 ] = [ω, iq s , ωids , 1]T and hd =
T ∞ ∞
[hd1 , hd2 , hd3 ]T = [ids , ωiq s , 1]T . k1 k6 δq σq2 dτ + δd k6 i2ds dτ ≤ 2V (0) (14)
Proof: The assumptions A3–5 imply that (4) holds with ξd∗ = 0 0

[k4 , −1, −dd ]T /k6 and where V (t) ≥ 0 is used. Because k1 k6 δq > 0 and δd k6 > 0, we
⎡ ⎤ can obtain
(k2 γq − k22 + k1 k5 )
∞
1 ⎢ k1 (k2 − γq + k4 ) ⎥
ξq∗ = ⎣ ⎦. (5) σq2 dτ < ∞ (15)
k1 k6 k1 0
(γq k3 TL − k2 k3 TL − k1 dq )
which implies that σq ∈ L2 . Since V̇ ≤ 0 as shown in (12),

we can see that V (t) is nonincreasing and is upper bounded as
V (t) ≤ V (0). This implies that σq ∈ L∞ , ids ∈ L∞ , ξq ∈ L∞ ,
III. ADAPTIVE CONTROLLER DESIGN AND
and ξd ∈ L∞ . Because of the transfer function from σq to ωe ,
STABILITY ANALYSIS
Hω e σ q (s) is strictly positive real, and σq ∈ L2 , we can use the
Theorem 1: Let the control input variables Vq s and Vds be result of [12] to state that ωe converges to zero.
given by the following adaptive control law: 

4 
3
Vq s = −δq σq + ξq i hq i , Vds = −δd ids + ξdi hdi (6) IV. DISCRETIZED CONTROLLER
i=1 i=1
We can rearrange Vq s and Vds as
t t
1 1
ξq i =− hq i σq dτ, ξdi =− hdi ids dτ (7) Vq s = uq s + uq d , Vds = uds + udd , (16)
φ qi 0 φ di 0

where ξq i and ξdi are estimates of ξq∗i and ξdi ∗

, σq = γq ωe + β, where uq s and uds are the static terms given by
φq i > 0, φdi > 0, δq > 0, and δd > 0. Then, ωe converges to
uq s (t) = −δq γq ωe , uds (t) = −δd ids (17)
zero, ξq i and ξdi are bounded.
Proof: Along the similar line of [12], we will show the and uq d (t) and udd (t) are the dynamic terms given by
stability of the adaptive system. Let us  define the Lya-
punov functional as V (t) = σq2 + i2ds + 4i=1 k1 k6 φq i ξ˜q2i + d  4 3
3 ˜2 ˜ ∗ ˜ ∗
uq d (t) = −δq ωe − hq i ξq i , udd (t) = − ξdi hdi .
i=1 k6 φdi ξdi , where ξq i = ξq i − ξq i and ξdi = ξdi − ξdi . Its dt i=1 i=1
time derivative along the error dynamics (3) is given by (18)

4 
3 For digital implementation under a sufficiently small sampling
2V̇ = σq σ̇q + ids i̇ds − k1 k6 φq i ξ˜q i ξ˙q i − k6 φdi ξ˜di ξ˙di . time T , the static terms uq s and uds at the sampling instant kT
i=1 i=1 can be straightforwardly set as
(8)
On the other hand, (3) implies that uq s (k) = −δq γq ωe (k), uds (k) = −δd ids (k) (19)

σ̇q = −k1 k6 (Vq f − Vq s ), i̇ds = −k6 (Vdf − Vds ). (9) whereas the dynamic terms uq d (k) and udd (k) cannot be
straightforwardly obtained. By using the relation ω̇ = β and
Lemma 1, (6), and (7) imply that
⎧ the previous result [11], we can compute the derivative term
⎪
⎪ 
4
β(k) by using the following recursive equation:
⎪ Vq s = −δq σq + Vq f −
⎪ ξ˜q i hq i
⎨
i=1 ρ 1
(10) β(k) = β(k − 1) + [ω(k) − ω(k − 1)] (20)
⎪
⎪ 
3
T +ρ T +ρ
⎪
⎪ ξ˜di hdi
⎩ Vds = −δd ids + Vdf −
i=1 where ρ is a sufficiently small filter time constant to limit the
susceptibility of the derivative term β = ω̇ to noise. By using
1 1 the relation (7) and the previous result [11], we can compute the
ξ˙q i = − hq i σq , ξ˙di = − hdi ids (11)
φ qi φ di integral terms ξq i (k) and ξdi (k) by using the following recursive
IEEE TRANSACTIONS ON POWER ELECTRONICS 5

Fig. 1. Block diagram of the proposed control algorithm.

Fig. 3. Simulation results under no parameter variation.

Fig. 4. Simulation results under 200% variations of some parameters (R s ,

L s , J , T L , and λm ).

Fig. 2. Block diagram of the proposed PMSM control system.

equation:
T
ξq i (k + 1) = ξq i (k) − hq i (k)[γq ωe (k) + β(k)] (21)
φq i
T
ξdi (k + 1) = ξdi (k) − hdi (k)ids (k). (22)
φdi
And, therefore, the dynamic terms uq d (k) and udd (k) can be
computed by the following equation:
⎧
⎪
⎪  4
⎪
⎪ qd
u (k) = −δ β(k) + ξq i (k)hq i (k)
⎨ q
i=1
(23)
⎪
⎪ 3
⎪
⎪
⎩ udd (k) = ξdi (k)hdi (k)
Fig. 5. Simulation results under speed reverse case.
i=1
6 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 1, JANUARY 2011

Fig. 6. Experimental results. (a) ω d and ω. (b) V q s and V d s . (c) iq s and id s . (d) V a n and ia .

where β(k), ξq i (k), and ξdi (k) are updated by the recursive where ξq i (k) and ξdi (k) are updated by the following recursive
equations (20)–(22). Fig. 1 shows overall block diagram of the equations:
proposed control algorithm. ⎧ 1
⎪
⎪ ξq i (k + 1) = ξq i (k) − hq i (k)ωe (k)
⎪
⎪ 100
⎪
⎪
⎨ 1
− hq i (k)[ω(k) − ω(k − 1)] (25)
V. SIMULATION AND EXPERIMENT ⎪ 10000T
⎪
⎪
⎪
⎪
For simulation and experiment, we consider a PMSM (1) ⎪
⎩ ξ (k + 1) = ξ (k) − 1 h (k)i (k)
with p = 12, Rs = 0.99[Ω], Ls = 5.82[mH], λm = 0.0791 di di
10000
di ds
[V·sec/rad], J = 0.00121[kg·m2 ], B = 0.0003[N· m·sec/rad].
With δq = 0.01, δd = 0.001, φq i = 2, φdi = 2, γq = 100, and hq (k) = [ω(k), iq s (k), ω(k)ids (k), 1]T , hd (k) = [ids (k),
and ρ = 0, we can obtain the following control law: ω(k)iq s (k), 1]T . Fig. 2 shows the overall block diagram of
the proposed PMSM control system. In simulations and ex-
periments, the switching frequency and the sampling frequency
⎧ 0.01 (1/T ) are selected as 5[kHz], and a space vector pulsewidth
⎪
⎪ Vq s (k) = −ωe (k) − [ω(k) − ω(k − 1)]
⎪
⎪ T modulation technique is adopted. Figs. 3 and 4 show the sim-
⎪
⎪
⎪
⎪ 
4 ulation results using Matlab/Simulink about two cases: no pa-
⎪
⎨ + ξq i (k)hq i (k) rameter variation and 200% variations of some parameters (Rs ,
(24) Ls , J, TL , and λm ). In both cases, the desired motor speed (ωd )
⎪
⎪
i=1
⎪
⎪
⎪
⎪ increases from 125.66 [rad/sec] to 251.32 [rad/sec] and then de-
⎪
⎪ 
3
⎪
⎩ Vds (k) = −0.001ids (k) + ξdi (k)hdi (k) creases from 251.32 [rad/sec] to 125.66 [rad/sec]. Fig. 3 shows
i=1
the simulation results (ωd , ω, Vq s , Vds , iq s , ids , Van , and ia )
IEEE TRANSACTIONS ON POWER ELECTRONICS 7

Fig. 7. Experimental results (speed reverse case). (a) ω d and ω. (b) V q s and V d s . (c) iq s and id s . (d) V a n and ia .

under no parameter variation. In Figs. 3 and 4, we can observe torque values. We also proved that the speed error of the closed-
that our adaptive speed controller is very robust to model pa- loop system converges to zero and the parameter adaptation
rameter and load-torque variations. Fig. 4 shows the simulation error signal is bounded. Via various simulation and experimen-
results under 200% variations of some parameters (Rs , Ls , J, tal results, it was clearly proven that the proposed adaptive speed
TL , and λm ). For further verification, we also consider the case regulator gives very remarkable speed-control performance with
when the desired speed (ωd ) is reversed. In this case, the desired no information on the PMSM parameter and load-torque values.
motor speed (ωd ) goes down from 125.66 [rad/sec] to –125.66
[rad/sec] and then goes up from –125.66 [rad/sec] to 125.66
REFERENCES
[rad/sec]. Fig. 5 shows the simulation results for that case. Fig. 6
shows the experimental results about motor speed, voltage, and [1] T.-H. Liu, H.-T. Pu, and C.-K. Lin, “Implementation of an adap-
tive position control system of a permanent-magnet synchronous mo-
current under the same condition as Fig. 3. Fig. 6(a) shows the tor and its application,” IET Electr. Power Appl., vol. 4, pp. 121–130,
desired speed (ωd ), measured speed (ω), and Fig. 6(b) shows 2010.
the q-axis and d-axis voltages (Vq s , Vds ), respectivley. Fig. 6(c) [2] W.-T. Su and C.-M. Liaw, “Adaptive positioning control for a LPMSM
drive based on adapted inverse model and robust disturbance observer,”
shows the measured q-axis current (iq s ) and d-axis current (ids ), IEEE Trans. Power Electron., vol. 21, no. 2, pp. 505–517, Mar. 2006.
and Fig. 6(d) shows the line to neutral voltage (Van ) and phase [3] Y. A.-R. I. Mohamed, “A hybrid-type variable-structure instantaneous
a current (ia ). Fig. 7 shows the experimental results about the torque control with a robust adaptive torque observer for a high-
performance direct-drive PMSM,” IEEE Trans. Ind. Electron., vol. 54,
speed reverse case. no. 5, pp. 2491–2499, Oct. 2007.
[4] K.-H. Kim, “Model reference adaptive control-based adaptive current con-
trol scheme of a PM synchronous motor with an improved servo perfor-
VI. CONCLUSION mance,” IET Electr. Power Appl., vol. 3, pp. 8–18, 2009.
[5] C.-K. Lin, T.-H. Liu, and S.-H. Yang, “Nonlinear position controller design
We proposed an adaptive speed-regulator design method for with input-output linearisation technique for an interior permanent magnet
a PMSM. The proposed adaptive speed controller is robust be- synchronous motor control system,” IET Power Electron., vol. 1, pp. 14–
cause it does not depend on the PMSM parameter and load- 26, 2008.
8 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 1, JANUARY 2011

[6] M. N. Uddin and M. A. Rahman, “High-speed control of IPMSM drives IEEE Trans. Ind. Electron., vol. 53, no. 4, pp. 1198–1208, Aug.
using improved fuzzy logic algorithms,” IEEE Trans. Ind. Electron., 2006.
vol. 54, no. 1, pp. 190–199, Feb. 2007. [10] S. Li and Z. Liu, “Adaptive speed control for permanent-magnet syn-
[7] Y.-S. Kung, C.-C. Huang, and M.-H. Tsai, “FPGA realization of an adap- chronous motor system with variations of load inertia,” IEEE Trans. Ind.
tive fuzzy controller for PMLSM drive,” IEEE Trans. Ind. Electron., Electron., vol. 56, no. 8, pp. 3050–3059, Aug. 2009.
vol. 56, no. 8, pp. 2923–2932, Aug. 2009. [11] K. J. Astrom and B. Witternmark, Computer-Controlled Systems - Theory
[8] K.-B. Lee and F. Blaabjerg, “Robust and stable disturbance observer of and Design. Englewood Cliffs, NJ: Prentice-Hall, 1990.
servo system for low-speed operation,” IEEE Trans. Ind. Appl., vol. 43, [12] F. L. Lewis, C. T. Abdallah, and D. M. Dawson, Control of Robot Manip-
no. 3, pp. 627–635, May/Jun. 2007. ulators. New York, NY: Macmillan, 1993.
[9] Y. Zhang, C. M. Akujuobi, W. H. Ali, C. L. Tolliver, and L.-S.
Shieh, “Load disturbance resistance speed controller design for PMSM,”